winevulkan: Update to VK spec version 1.2.195.

[wine.git] / dlls / cabinet / deflate.c

/* deflate.c -- compress data using the deflation algorithm
 *
 * Copyright (C) 1995-2017 Jean-loup Gailly and Mark Adler
 *
 * This software is provided 'as-is', without any express or implied
 * warranty.  In no event will the authors be held liable for any damages
 * arising from the use of this software.
 *
 * Permission is granted to anyone to use this software for any purpose,
 * including commercial applications, and to alter it and redistribute it
 * freely, subject to the following restrictions:
 *
 * 1. The origin of this software must not be misrepresented; you must not
 *    claim that you wrote the original software. If you use this software
 *    in a product, an acknowledgment in the product documentation would be
 *    appreciated but is not required.
 * 2. Altered source versions must be plainly marked as such, and must not be
 *    misrepresented as being the original software.
 * 3. This notice may not be removed or altered from any source distribution.
 *
 * Jean-loup Gailly        Mark Adler
 * jloup@gzip.org          madler@alumni.caltech.edu
 */

#include <stdlib.h>
#include <string.h>
#include <limits.h>
#include "zlib.h"

#define DEF_MEM_LEVEL 8
#define DEF_WBITS MAX_WBITS
#define zmemcpy memcpy
#define zmemzero(dest, len) memset(dest, 0, len)

#define Assert(cond,msg)
#define Trace(x)
#define Tracev(x)
#define Tracevv(x)
#define Tracecv(c,x)

#define ZALLOC(strm, items, size) \
           (*((strm)->zalloc))((strm)->opaque, (items), (size))
#define ZFREE(strm, addr)  (*((strm)->zfree))((strm)->opaque, (voidpf)(addr))
#define TRY_FREE(s, p) {if (p) ZFREE(s, p);}

/* Reverse the bytes in a 32-bit value */
#define ZSWAP32(q) ((((q) >> 24) & 0xff) + (((q) >> 8) & 0xff00) + \
                    (((q) & 0xff00) << 8) + (((q) & 0xff) << 24))

static const char * const z_errmsg[10] = {
    (z_const char *)"need dictionary",     /* Z_NEED_DICT       2  */
    (z_const char *)"stream end",          /* Z_STREAM_END      1  */
    (z_const char *)"",                    /* Z_OK              0  */
    (z_const char *)"file error",          /* Z_ERRNO         (-1) */
    (z_const char *)"stream error",        /* Z_STREAM_ERROR  (-2) */
    (z_const char *)"data error",          /* Z_DATA_ERROR    (-3) */
    (z_const char *)"insufficient memory", /* Z_MEM_ERROR     (-4) */
    (z_const char *)"buffer error",        /* Z_BUF_ERROR     (-5) */
    (z_const char *)"incompatible version",/* Z_VERSION_ERROR (-6) */
    (z_const char *)""
};

#define ERR_MSG(err) z_errmsg[Z_NEED_DICT-(err)]

#define ERR_RETURN(strm,err) \
  return (strm->msg = ERR_MSG(err), (err))
/* To be used only when the state is known to be valid */

#define STORED_BLOCK 0
#define STATIC_TREES 1
#define DYN_TREES    2
/* The three kinds of block type */

#define MIN_MATCH  3
#define MAX_MATCH  258
/* The minimum and maximum match lengths */

#define PRESET_DICT 0x20 /* preset dictionary flag in zlib header */

#define BASE 65521U     /* largest prime smaller than 65536 */
#define NMAX 5552
/* NMAX is the largest n such that 255n(n+1)/2 + (n+1)(BASE-1) <= 2^32-1 */

#define DO1(buf,i)  {adler += (buf)[i]; sum2 += adler;}
#define DO2(buf,i)  DO1(buf,i); DO1(buf,i+1);
#define DO4(buf,i)  DO2(buf,i); DO2(buf,i+2);
#define DO8(buf,i)  DO4(buf,i); DO4(buf,i+4);
#define DO16(buf)   DO8(buf,0); DO8(buf,8);

#define MOD(a) a %= BASE
#define MOD28(a) a %= BASE
#define MOD63(a) a %= BASE

static uLong adler32( uLong adler, const Bytef *buf, uInt len )
{
    unsigned long sum2;
    unsigned n;

    /* split Adler-32 into component sums */
    sum2 = (adler >> 16) & 0xffff;
    adler &= 0xffff;

    /* in case user likes doing a byte at a time, keep it fast */
    if (len == 1) {
        adler += buf[0];
        if (adler >= BASE)
            adler -= BASE;
        sum2 += adler;
        if (sum2 >= BASE)
            sum2 -= BASE;
        return adler | (sum2 << 16);
    }

    /* initial Adler-32 value (deferred check for len == 1 speed) */
    if (buf == Z_NULL)
        return 1L;

    /* in case short lengths are provided, keep it somewhat fast */
    if (len < 16) {
        while (len--) {
            adler += *buf++;
            sum2 += adler;
        }
        if (adler >= BASE)
            adler -= BASE;
        MOD28(sum2);            /* only added so many BASE's */
        return adler | (sum2 << 16);
    }

    /* do length NMAX blocks -- requires just one modulo operation */
    while (len >= NMAX) {
        len -= NMAX;
        n = NMAX / 16;          /* NMAX is divisible by 16 */
        do {
            DO16(buf);          /* 16 sums unrolled */
            buf += 16;
        } while (--n);
        MOD(adler);
        MOD(sum2);
    }

    /* do remaining bytes (less than NMAX, still just one modulo) */
    if (len) {                  /* avoid modulos if none remaining */
        while (len >= 16) {
            len -= 16;
            DO16(buf);
            buf += 16;
        }
        while (len--) {
            adler += *buf++;
            sum2 += adler;
        }
        MOD(adler);
        MOD(sum2);
    }

    /* return recombined sums */
    return adler | (sum2 << 16);
}

/* ===========================================================================
 * Internal compression state.
 */

#define LENGTH_CODES 29
/* number of length codes, not counting the special END_BLOCK code */

#define LITERALS  256
/* number of literal bytes 0..255 */

#define L_CODES (LITERALS+1+LENGTH_CODES)
/* number of Literal or Length codes, including the END_BLOCK code */

#define D_CODES   30
/* number of distance codes */

#define BL_CODES  19
/* number of codes used to transfer the bit lengths */

#define HEAP_SIZE (2*L_CODES+1)
/* maximum heap size */

#define MAX_BITS 15
/* All codes must not exceed MAX_BITS bits */

#define Buf_size 16
/* size of bit buffer in bi_buf */

#define INIT_STATE    42    /* zlib header -> BUSY_STATE */
#ifdef GZIP
#  define GZIP_STATE  57    /* gzip header -> BUSY_STATE | EXTRA_STATE */
#endif
#define EXTRA_STATE   69    /* gzip extra block -> NAME_STATE */
#define NAME_STATE    73    /* gzip file name -> COMMENT_STATE */
#define COMMENT_STATE 91    /* gzip comment -> HCRC_STATE */
#define HCRC_STATE   103    /* gzip header CRC -> BUSY_STATE */
#define BUSY_STATE   113    /* deflate -> FINISH_STATE */
#define FINISH_STATE 666    /* stream complete */
/* Stream status */


/* Data structure describing a single value and its code string. */
typedef struct ct_data_s {
    union {
        ush  freq;       /* frequency count */
        ush  code;       /* bit string */
    } fc;
    union {
        ush  dad;        /* father node in Huffman tree */
        ush  len;        /* length of bit string */
    } dl;
} FAR ct_data;

#define Freq fc.freq
#define Code fc.code
#define Dad  dl.dad
#define Len  dl.len

typedef struct static_tree_desc_s  static_tree_desc;

typedef struct tree_desc_s {
    ct_data *dyn_tree;           /* the dynamic tree */
    int     max_code;            /* largest code with non zero frequency */
    const static_tree_desc *stat_desc;  /* the corresponding static tree */
} FAR tree_desc;

typedef ush Pos;
typedef Pos FAR Posf;
typedef unsigned IPos;

/* A Pos is an index in the character window. We use short instead of int to
 * save space in the various tables. IPos is used only for parameter passing.
 */

typedef struct internal_state {
    z_streamp strm;      /* pointer back to this zlib stream */
    int   status;        /* as the name implies */
    Bytef *pending_buf;  /* output still pending */
    ulg   pending_buf_size; /* size of pending_buf */
    Bytef *pending_out;  /* next pending byte to output to the stream */
    ulg   pending;       /* nb of bytes in the pending buffer */
    int   wrap;          /* bit 0 true for zlib, bit 1 true for gzip */
    gz_headerp  gzhead;  /* gzip header information to write */
    ulg   gzindex;       /* where in extra, name, or comment */
    Byte  method;        /* can only be DEFLATED */
    int   last_flush;    /* value of flush param for previous deflate call */

                /* used by deflate.c: */

    uInt  w_size;        /* LZ77 window size (32K by default) */
    uInt  w_bits;        /* log2(w_size)  (8..16) */
    uInt  w_mask;        /* w_size - 1 */

    Bytef *window;
    /* Sliding window. Input bytes are read into the second half of the window,
     * and move to the first half later to keep a dictionary of at least wSize
     * bytes. With this organization, matches are limited to a distance of
     * wSize-MAX_MATCH bytes, but this ensures that IO is always
     * performed with a length multiple of the block size. Also, it limits
     * the window size to 64K, which is quite useful on MSDOS.
     * To do: use the user input buffer as sliding window.
     */

    ulg window_size;
    /* Actual size of window: 2*wSize, except when the user input buffer
     * is directly used as sliding window.
     */

    Posf *prev;
    /* Link to older string with same hash index. To limit the size of this
     * array to 64K, this link is maintained only for the last 32K strings.
     * An index in this array is thus a window index modulo 32K.
     */

    Posf *head; /* Heads of the hash chains or NIL. */

    uInt  ins_h;          /* hash index of string to be inserted */
    uInt  hash_size;      /* number of elements in hash table */
    uInt  hash_bits;      /* log2(hash_size) */
    uInt  hash_mask;      /* hash_size-1 */

    uInt  hash_shift;
    /* Number of bits by which ins_h must be shifted at each input
     * step. It must be such that after MIN_MATCH steps, the oldest
     * byte no longer takes part in the hash key, that is:
     *   hash_shift * MIN_MATCH >= hash_bits
     */

    long block_start;
    /* Window position at the beginning of the current output block. Gets
     * negative when the window is moved backwards.
     */

    uInt match_length;           /* length of best match */
    IPos prev_match;             /* previous match */
    int match_available;         /* set if previous match exists */
    uInt strstart;               /* start of string to insert */
    uInt match_start;            /* start of matching string */
    uInt lookahead;              /* number of valid bytes ahead in window */

    uInt prev_length;
    /* Length of the best match at previous step. Matches not greater than this
     * are discarded. This is used in the lazy match evaluation.
     */

    uInt max_chain_length;
    /* To speed up deflation, hash chains are never searched beyond this
     * length.  A higher limit improves compression ratio but degrades the
     * speed.
     */

    uInt max_lazy_match;
    /* Attempt to find a better match only when the current match is strictly
     * smaller than this value. This mechanism is used only for compression
     * levels >= 4.
     */
#   define max_insert_length  max_lazy_match
    /* Insert new strings in the hash table only if the match length is not
     * greater than this length. This saves time but degrades compression.
     * max_insert_length is used only for compression levels <= 3.
     */

    int level;    /* compression level (1..9) */
    int strategy; /* favor or force Huffman coding*/

    uInt good_match;
    /* Use a faster search when the previous match is longer than this */

    int nice_match; /* Stop searching when current match exceeds this */

                /* used by trees.c: */
    /* Didn't use ct_data typedef below to suppress compiler warning */
    struct ct_data_s dyn_ltree[HEAP_SIZE];   /* literal and length tree */
    struct ct_data_s dyn_dtree[2*D_CODES+1]; /* distance tree */
    struct ct_data_s bl_tree[2*BL_CODES+1];  /* Huffman tree for bit lengths */

    struct tree_desc_s l_desc;               /* desc. for literal tree */
    struct tree_desc_s d_desc;               /* desc. for distance tree */
    struct tree_desc_s bl_desc;              /* desc. for bit length tree */

    ush bl_count[MAX_BITS+1];
    /* number of codes at each bit length for an optimal tree */

    int heap[2*L_CODES+1];      /* heap used to build the Huffman trees */
    int heap_len;               /* number of elements in the heap */
    int heap_max;               /* element of largest frequency */
    /* The sons of heap[n] are heap[2*n] and heap[2*n+1]. heap[0] is not used.
     * The same heap array is used to build all trees.
     */

    uch depth[2*L_CODES+1];
    /* Depth of each subtree used as tie breaker for trees of equal frequency
     */

    uchf *l_buf;          /* buffer for literals or lengths */

    uInt  lit_bufsize;
    /* Size of match buffer for literals/lengths.  There are 4 reasons for
     * limiting lit_bufsize to 64K:
     *   - frequencies can be kept in 16 bit counters
     *   - if compression is not successful for the first block, all input
     *     data is still in the window so we can still emit a stored block even
     *     when input comes from standard input.  (This can also be done for
     *     all blocks if lit_bufsize is not greater than 32K.)
     *   - if compression is not successful for a file smaller than 64K, we can
     *     even emit a stored file instead of a stored block (saving 5 bytes).
     *     This is applicable only for zip (not gzip or zlib).
     *   - creating new Huffman trees less frequently may not provide fast
     *     adaptation to changes in the input data statistics. (Take for
     *     example a binary file with poorly compressible code followed by
     *     a highly compressible string table.) Smaller buffer sizes give
     *     fast adaptation but have of course the overhead of transmitting
     *     trees more frequently.
     *   - I can't count above 4
     */

    uInt last_lit;      /* running index in l_buf */

    ushf *d_buf;
    /* Buffer for distances. To simplify the code, d_buf and l_buf have
     * the same number of elements. To use different lengths, an extra flag
     * array would be necessary.
     */

    ulg opt_len;        /* bit length of current block with optimal trees */
    ulg static_len;     /* bit length of current block with static trees */
    uInt matches;       /* number of string matches in current block */
    uInt insert;        /* bytes at end of window left to insert */

#ifdef ZLIB_DEBUG
    ulg compressed_len; /* total bit length of compressed file mod 2^32 */
    ulg bits_sent;      /* bit length of compressed data sent mod 2^32 */
#endif

    ush bi_buf;
    /* Output buffer. bits are inserted starting at the bottom (least
     * significant bits).
     */
    int bi_valid;
    /* Number of valid bits in bi_buf.  All bits above the last valid bit
     * are always zero.
     */

    ulg high_water;
    /* High water mark offset in window for initialized bytes -- bytes above
     * this are set to zero in order to avoid memory check warnings when
     * longest match routines access bytes past the input.  This is then
     * updated to the new high water mark.
     */

} FAR deflate_state;

/* Output a byte on the stream.
 * IN assertion: there is enough room in pending_buf.
 */
#define put_byte(s, c) {s->pending_buf[s->pending++] = (Bytef)(c);}


#define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
/* Minimum amount of lookahead, except at the end of the input file.
 * See deflate.c for comments about the MIN_MATCH+1.
 */

#define MAX_DIST(s)  ((s)->w_size-MIN_LOOKAHEAD)
/* In order to simplify the code, particularly on 16 bit machines, match
 * distances are limited to MAX_DIST instead of WSIZE.
 */

#define WIN_INIT MAX_MATCH
/* Number of bytes after end of data in window to initialize in order to avoid
   memory checker errors from longest match routines */


#define MAX_BL_BITS 7
/* Bit length codes must not exceed MAX_BL_BITS bits */

#define END_BLOCK 256
/* end of block literal code */

#define REP_3_6      16
/* repeat previous bit length 3-6 times (2 bits of repeat count) */

#define REPZ_3_10    17
/* repeat a zero length 3-10 times  (3 bits of repeat count) */

#define REPZ_11_138  18
/* repeat a zero length 11-138 times  (7 bits of repeat count) */

static const int extra_lbits[LENGTH_CODES] /* extra bits for each length code */
   = {0,0,0,0,0,0,0,0,1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4,5,5,5,5,0};

static const int extra_dbits[D_CODES] /* extra bits for each distance code */
   = {0,0,0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13};

static const int extra_blbits[BL_CODES]/* extra bits for each bit length code */
   = {0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,2,3,7};

static const uch bl_order[BL_CODES]
   = {16,17,18,0,8,7,9,6,10,5,11,4,12,3,13,2,14,1,15};
/* The lengths of the bit length codes are sent in order of decreasing
 * probability, to avoid transmitting the lengths for unused bit length codes.
 */

/* ===========================================================================
 * Local data. These are initialized only once.
 */

#define DIST_CODE_LEN  512 /* see definition of array dist_code below */

static const ct_data static_ltree[L_CODES+2] = {
{{ 12},{  8}}, {{140},{  8}}, {{ 76},{  8}}, {{204},{  8}}, {{ 44},{  8}},
{{172},{  8}}, {{108},{  8}}, {{236},{  8}}, {{ 28},{  8}}, {{156},{  8}},
{{ 92},{  8}}, {{220},{  8}}, {{ 60},{  8}}, {{188},{  8}}, {{124},{  8}},
{{252},{  8}}, {{  2},{  8}}, {{130},{  8}}, {{ 66},{  8}}, {{194},{  8}},
{{ 34},{  8}}, {{162},{  8}}, {{ 98},{  8}}, {{226},{  8}}, {{ 18},{  8}},
{{146},{  8}}, {{ 82},{  8}}, {{210},{  8}}, {{ 50},{  8}}, {{178},{  8}},
{{114},{  8}}, {{242},{  8}}, {{ 10},{  8}}, {{138},{  8}}, {{ 74},{  8}},
{{202},{  8}}, {{ 42},{  8}}, {{170},{  8}}, {{106},{  8}}, {{234},{  8}},
{{ 26},{  8}}, {{154},{  8}}, {{ 90},{  8}}, {{218},{  8}}, {{ 58},{  8}},
{{186},{  8}}, {{122},{  8}}, {{250},{  8}}, {{  6},{  8}}, {{134},{  8}},
{{ 70},{  8}}, {{198},{  8}}, {{ 38},{  8}}, {{166},{  8}}, {{102},{  8}},
{{230},{  8}}, {{ 22},{  8}}, {{150},{  8}}, {{ 86},{  8}}, {{214},{  8}},
{{ 54},{  8}}, {{182},{  8}}, {{118},{  8}}, {{246},{  8}}, {{ 14},{  8}},
{{142},{  8}}, {{ 78},{  8}}, {{206},{  8}}, {{ 46},{  8}}, {{174},{  8}},
{{110},{  8}}, {{238},{  8}}, {{ 30},{  8}}, {{158},{  8}}, {{ 94},{  8}},
{{222},{  8}}, {{ 62},{  8}}, {{190},{  8}}, {{126},{  8}}, {{254},{  8}},
{{  1},{  8}}, {{129},{  8}}, {{ 65},{  8}}, {{193},{  8}}, {{ 33},{  8}},
{{161},{  8}}, {{ 97},{  8}}, {{225},{  8}}, {{ 17},{  8}}, {{145},{  8}},
{{ 81},{  8}}, {{209},{  8}}, {{ 49},{  8}}, {{177},{  8}}, {{113},{  8}},
{{241},{  8}}, {{  9},{  8}}, {{137},{  8}}, {{ 73},{  8}}, {{201},{  8}},
{{ 41},{  8}}, {{169},{  8}}, {{105},{  8}}, {{233},{  8}}, {{ 25},{  8}},
{{153},{  8}}, {{ 89},{  8}}, {{217},{  8}}, {{ 57},{  8}}, {{185},{  8}},
{{121},{  8}}, {{249},{  8}}, {{  5},{  8}}, {{133},{  8}}, {{ 69},{  8}},
{{197},{  8}}, {{ 37},{  8}}, {{165},{  8}}, {{101},{  8}}, {{229},{  8}},
{{ 21},{  8}}, {{149},{  8}}, {{ 85},{  8}}, {{213},{  8}}, {{ 53},{  8}},
{{181},{  8}}, {{117},{  8}}, {{245},{  8}}, {{ 13},{  8}}, {{141},{  8}},
{{ 77},{  8}}, {{205},{  8}}, {{ 45},{  8}}, {{173},{  8}}, {{109},{  8}},
{{237},{  8}}, {{ 29},{  8}}, {{157},{  8}}, {{ 93},{  8}}, {{221},{  8}},
{{ 61},{  8}}, {{189},{  8}}, {{125},{  8}}, {{253},{  8}}, {{ 19},{  9}},
{{275},{  9}}, {{147},{  9}}, {{403},{  9}}, {{ 83},{  9}}, {{339},{  9}},
{{211},{  9}}, {{467},{  9}}, {{ 51},{  9}}, {{307},{  9}}, {{179},{  9}},
{{435},{  9}}, {{115},{  9}}, {{371},{  9}}, {{243},{  9}}, {{499},{  9}},
{{ 11},{  9}}, {{267},{  9}}, {{139},{  9}}, {{395},{  9}}, {{ 75},{  9}},
{{331},{  9}}, {{203},{  9}}, {{459},{  9}}, {{ 43},{  9}}, {{299},{  9}},
{{171},{  9}}, {{427},{  9}}, {{107},{  9}}, {{363},{  9}}, {{235},{  9}},
{{491},{  9}}, {{ 27},{  9}}, {{283},{  9}}, {{155},{  9}}, {{411},{  9}},
{{ 91},{  9}}, {{347},{  9}}, {{219},{  9}}, {{475},{  9}}, {{ 59},{  9}},
{{315},{  9}}, {{187},{  9}}, {{443},{  9}}, {{123},{  9}}, {{379},{  9}},
{{251},{  9}}, {{507},{  9}}, {{  7},{  9}}, {{263},{  9}}, {{135},{  9}},
{{391},{  9}}, {{ 71},{  9}}, {{327},{  9}}, {{199},{  9}}, {{455},{  9}},
{{ 39},{  9}}, {{295},{  9}}, {{167},{  9}}, {{423},{  9}}, {{103},{  9}},
{{359},{  9}}, {{231},{  9}}, {{487},{  9}}, {{ 23},{  9}}, {{279},{  9}},
{{151},{  9}}, {{407},{  9}}, {{ 87},{  9}}, {{343},{  9}}, {{215},{  9}},
{{471},{  9}}, {{ 55},{  9}}, {{311},{  9}}, {{183},{  9}}, {{439},{  9}},
{{119},{  9}}, {{375},{  9}}, {{247},{  9}}, {{503},{  9}}, {{ 15},{  9}},
{{271},{  9}}, {{143},{  9}}, {{399},{  9}}, {{ 79},{  9}}, {{335},{  9}},
{{207},{  9}}, {{463},{  9}}, {{ 47},{  9}}, {{303},{  9}}, {{175},{  9}},
{{431},{  9}}, {{111},{  9}}, {{367},{  9}}, {{239},{  9}}, {{495},{  9}},
{{ 31},{  9}}, {{287},{  9}}, {{159},{  9}}, {{415},{  9}}, {{ 95},{  9}},
{{351},{  9}}, {{223},{  9}}, {{479},{  9}}, {{ 63},{  9}}, {{319},{  9}},
{{191},{  9}}, {{447},{  9}}, {{127},{  9}}, {{383},{  9}}, {{255},{  9}},
{{511},{  9}}, {{  0},{  7}}, {{ 64},{  7}}, {{ 32},{  7}}, {{ 96},{  7}},
{{ 16},{  7}}, {{ 80},{  7}}, {{ 48},{  7}}, {{112},{  7}}, {{  8},{  7}},
{{ 72},{  7}}, {{ 40},{  7}}, {{104},{  7}}, {{ 24},{  7}}, {{ 88},{  7}},
{{ 56},{  7}}, {{120},{  7}}, {{  4},{  7}}, {{ 68},{  7}}, {{ 36},{  7}},
{{100},{  7}}, {{ 20},{  7}}, {{ 84},{  7}}, {{ 52},{  7}}, {{116},{  7}},
{{  3},{  8}}, {{131},{  8}}, {{ 67},{  8}}, {{195},{  8}}, {{ 35},{  8}},
{{163},{  8}}, {{ 99},{  8}}, {{227},{  8}}
};

static const ct_data static_dtree[D_CODES] = {
{{ 0},{ 5}}, {{16},{ 5}}, {{ 8},{ 5}}, {{24},{ 5}}, {{ 4},{ 5}},
{{20},{ 5}}, {{12},{ 5}}, {{28},{ 5}}, {{ 2},{ 5}}, {{18},{ 5}},
{{10},{ 5}}, {{26},{ 5}}, {{ 6},{ 5}}, {{22},{ 5}}, {{14},{ 5}},
{{30},{ 5}}, {{ 1},{ 5}}, {{17},{ 5}}, {{ 9},{ 5}}, {{25},{ 5}},
{{ 5},{ 5}}, {{21},{ 5}}, {{13},{ 5}}, {{29},{ 5}}, {{ 3},{ 5}},
{{19},{ 5}}, {{11},{ 5}}, {{27},{ 5}}, {{ 7},{ 5}}, {{23},{ 5}}
};

static const uch _dist_code[DIST_CODE_LEN] = {
 0,  1,  2,  3,  4,  4,  5,  5,  6,  6,  6,  6,  7,  7,  7,  7,  8,  8,  8,  8,
 8,  8,  8,  8,  9,  9,  9,  9,  9,  9,  9,  9, 10, 10, 10, 10, 10, 10, 10, 10,
10, 10, 10, 10, 10, 10, 10, 10, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
11, 11, 11, 11, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12,
12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 13, 13, 13, 13,
13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13,
13, 13, 13, 13, 13, 13, 13, 13, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14,
14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14,
14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14,
14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 15, 15, 15, 15, 15, 15, 15, 15,
15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,  0,  0, 16, 17,
18, 18, 19, 19, 20, 20, 20, 20, 21, 21, 21, 21, 22, 22, 22, 22, 22, 22, 22, 22,
23, 23, 23, 23, 23, 23, 23, 23, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
24, 24, 24, 24, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25,
26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26,
26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 27, 27, 27, 27, 27, 27, 27, 27,
27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27,
27, 27, 27, 27, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28,
28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28,
28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28,
28, 28, 28, 28, 28, 28, 28, 28, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29,
29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29,
29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29,
29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29
};

static const uch _length_code[MAX_MATCH-MIN_MATCH+1]= {
 0,  1,  2,  3,  4,  5,  6,  7,  8,  8,  9,  9, 10, 10, 11, 11, 12, 12, 12, 12,
13, 13, 13, 13, 14, 14, 14, 14, 15, 15, 15, 15, 16, 16, 16, 16, 16, 16, 16, 16,
17, 17, 17, 17, 17, 17, 17, 17, 18, 18, 18, 18, 18, 18, 18, 18, 19, 19, 19, 19,
19, 19, 19, 19, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20,
21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 22, 22, 22, 22,
22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 23, 23, 23, 23, 23, 23, 23, 23,
23, 23, 23, 23, 23, 23, 23, 23, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25,
25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 26, 26, 26, 26, 26, 26, 26, 26,
26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26,
26, 26, 26, 26, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27,
27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 28
};

static const int base_length[LENGTH_CODES] = {
0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, 28, 32, 40, 48, 56,
64, 80, 96, 112, 128, 160, 192, 224, 0
};

static const int base_dist[D_CODES] = {
    0,     1,     2,     3,     4,     6,     8,    12,    16,    24,
   32,    48,    64,    96,   128,   192,   256,   384,   512,   768,
 1024,  1536,  2048,  3072,  4096,  6144,  8192, 12288, 16384, 24576
};


struct static_tree_desc_s {
    const ct_data *static_tree;  /* static tree or NULL */
    const intf *extra_bits;      /* extra bits for each code or NULL */
    int     extra_base;          /* base index for extra_bits */
    int     elems;               /* max number of elements in the tree */
    int     max_length;          /* max bit length for the codes */
};

static const static_tree_desc  static_l_desc =
{static_ltree, extra_lbits, LITERALS+1, L_CODES, MAX_BITS};

static const static_tree_desc  static_d_desc =
{static_dtree, extra_dbits, 0,          D_CODES, MAX_BITS};

static const static_tree_desc  static_bl_desc =
{(const ct_data *)0, extra_blbits, 0,   BL_CODES, MAX_BL_BITS};

#define send_code(s, c, tree) send_bits(s, tree[c].Code, tree[c].Len)
   /* Send a code of the given tree. c and tree must not have side effects */

#define d_code(dist) \
   ((dist) < 256 ? _dist_code[dist] : _dist_code[256+((dist)>>7)])
/* Mapping from a distance to a distance code. dist is the distance - 1 and
 * must not have side effects. _dist_code[256] and _dist_code[257] are never
 * used.
 */

# define _tr_tally_lit(s, c, flush) \
  { uch cc = (c); \
    s->d_buf[s->last_lit] = 0; \
    s->l_buf[s->last_lit++] = cc; \
    s->dyn_ltree[cc].Freq++; \
    flush = (s->last_lit == s->lit_bufsize-1); \
   }
# define _tr_tally_dist(s, distance, length, flush) \
  { uch len = (uch)(length); \
    ush dist = (ush)(distance); \
    s->d_buf[s->last_lit] = dist; \
    s->l_buf[s->last_lit++] = len; \
    dist--; \
    s->dyn_ltree[_length_code[len]+LITERALS+1].Freq++; \
    s->dyn_dtree[d_code(dist)].Freq++; \
    flush = (s->last_lit == s->lit_bufsize-1); \
  }


/* ===========================================================================
 * Output a short LSB first on the stream.
 * IN assertion: there is enough room in pendingBuf.
 */
#define put_short(s, w) { \
    put_byte(s, (uch)((w) & 0xff)); \
    put_byte(s, (uch)((ush)(w) >> 8)); \
}

#define send_bits(s, value, length) \
{ int len = length;\
  if (s->bi_valid > (int)Buf_size - len) {\
    int val = (int)value;\
    s->bi_buf |= (ush)val << s->bi_valid;\
    put_short(s, s->bi_buf);\
    s->bi_buf = (ush)val >> (Buf_size - s->bi_valid);\
    s->bi_valid += len - Buf_size;\
  } else {\
    s->bi_buf |= (ush)(value) << s->bi_valid;\
    s->bi_valid += len;\
  }\
}


/* ===========================================================================
 * Send the block data compressed using the given Huffman trees
 */
static void compress_block( deflate_state *s, const ct_data *ltree, const ct_data *dtree )
{
    unsigned dist;      /* distance of matched string */
    int lc;             /* match length or unmatched char (if dist == 0) */
    unsigned lx = 0;    /* running index in l_buf */
    unsigned code;      /* the code to send */
    int extra;          /* number of extra bits to send */

    if (s->last_lit != 0) do {
        dist = s->d_buf[lx];
        lc = s->l_buf[lx++];
        if (dist == 0) {
            send_code(s, lc, ltree); /* send a literal byte */
            Tracecv(isgraph(lc), (stderr," '%c' ", lc));
        } else {
            /* Here, lc is the match length - MIN_MATCH */
            code = _length_code[lc];
            send_code(s, code+LITERALS+1, ltree); /* send the length code */
            extra = extra_lbits[code];
            if (extra != 0) {
                lc -= base_length[code];
                send_bits(s, lc, extra);       /* send the extra length bits */
            }
            dist--; /* dist is now the match distance - 1 */
            code = d_code(dist);
            Assert (code < D_CODES, "bad d_code");

            send_code(s, code, dtree);       /* send the distance code */
            extra = extra_dbits[code];
            if (extra != 0) {
                dist -= (unsigned)base_dist[code];
                send_bits(s, dist, extra);   /* send the extra distance bits */
            }
        } /* literal or match pair ? */

        /* Check that the overlay between pending_buf and d_buf+l_buf is ok: */
        Assert((uInt)(s->pending) < s->lit_bufsize + 2*lx,
               "pendingBuf overflow");

    } while (lx < s->last_lit);

    send_code(s, END_BLOCK, ltree);
}

/* ===========================================================================
 * Check if the data type is TEXT or BINARY, using the following algorithm:
 * - TEXT if the two conditions below are satisfied:
 *    a) There are no non-portable control characters belonging to the
 *       "black list" (0..6, 14..25, 28..31).
 *    b) There is at least one printable character belonging to the
 *       "white list" (9 {TAB}, 10 {LF}, 13 {CR}, 32..255).
 * - BINARY otherwise.
 * - The following partially-portable control characters form a
 *   "gray list" that is ignored in this detection algorithm:
 *   (7 {BEL}, 8 {BS}, 11 {VT}, 12 {FF}, 26 {SUB}, 27 {ESC}).
 * IN assertion: the fields Freq of dyn_ltree are set.
 */
static int detect_data_type( deflate_state *s )
{
    /* black_mask is the bit mask of black-listed bytes
     * set bits 0..6, 14..25, and 28..31
     * 0xf3ffc07f = binary 11110011111111111100000001111111
     */
    unsigned long black_mask = 0xf3ffc07fUL;
    int n;

    /* Check for non-textual ("black-listed") bytes. */
    for (n = 0; n <= 31; n++, black_mask >>= 1)
        if ((black_mask & 1) && (s->dyn_ltree[n].Freq != 0))
            return Z_BINARY;

    /* Check for textual ("white-listed") bytes. */
    if (s->dyn_ltree[9].Freq != 0 || s->dyn_ltree[10].Freq != 0
            || s->dyn_ltree[13].Freq != 0)
        return Z_TEXT;
    for (n = 32; n < LITERALS; n++)
        if (s->dyn_ltree[n].Freq != 0)
            return Z_TEXT;

    /* There are no "black-listed" or "white-listed" bytes:
     * this stream either is empty or has tolerated ("gray-listed") bytes only.
     */
    return Z_BINARY;
}

/* ===========================================================================
 * Reverse the first len bits of a code, using straightforward code (a faster
 * method would use a table)
 * IN assertion: 1 <= len <= 15
 */
static unsigned bi_reverse( unsigned code, int len )
{
    register unsigned res = 0;
    do {
        res |= code & 1;
        code >>= 1, res <<= 1;
    } while (--len > 0);
    return res >> 1;
}

/* ===========================================================================
 * Flush the bit buffer, keeping at most 7 bits in it.
 */
static void bi_flush( deflate_state *s )
{
    if (s->bi_valid == 16) {
        put_short(s, s->bi_buf);
        s->bi_buf = 0;
        s->bi_valid = 0;
    } else if (s->bi_valid >= 8) {
        put_byte(s, (Byte)s->bi_buf);
        s->bi_buf >>= 8;
        s->bi_valid -= 8;
    }
}

/* ===========================================================================
 * Flush the bit buffer and align the output on a byte boundary
 */
static void bi_windup( deflate_state *s )
{
    if (s->bi_valid > 8) {
        put_short(s, s->bi_buf);
    } else if (s->bi_valid > 0) {
        put_byte(s, (Byte)s->bi_buf);
    }
    s->bi_buf = 0;
    s->bi_valid = 0;
#ifdef ZLIB_DEBUG
    s->bits_sent = (s->bits_sent+7) & ~7;
#endif
}

/* ===========================================================================
 * Initialize a new block.
 */
static void init_block( deflate_state *s )
{
    int n; /* iterates over tree elements */

    /* Initialize the trees. */
    for (n = 0; n < L_CODES;  n++) s->dyn_ltree[n].Freq = 0;
    for (n = 0; n < D_CODES;  n++) s->dyn_dtree[n].Freq = 0;
    for (n = 0; n < BL_CODES; n++) s->bl_tree[n].Freq = 0;

    s->dyn_ltree[END_BLOCK].Freq = 1;
    s->opt_len = s->static_len = 0L;
    s->last_lit = s->matches = 0;
}

/* ===========================================================================
 * Initialize the tree data structures for a new zlib stream.
 */
static void _tr_init( deflate_state *s )
{
    s->l_desc.dyn_tree = s->dyn_ltree;
    s->l_desc.stat_desc = &static_l_desc;

    s->d_desc.dyn_tree = s->dyn_dtree;
    s->d_desc.stat_desc = &static_d_desc;

    s->bl_desc.dyn_tree = s->bl_tree;
    s->bl_desc.stat_desc = &static_bl_desc;

    s->bi_buf = 0;
    s->bi_valid = 0;
#ifdef ZLIB_DEBUG
    s->compressed_len = 0L;
    s->bits_sent = 0L;
#endif

    /* Initialize the first block of the first file: */
    init_block(s);
}

#define SMALLEST 1
/* Index within the heap array of least frequent node in the Huffman tree */


/* ===========================================================================
 * Remove the smallest element from the heap and recreate the heap with
 * one less element. Updates heap and heap_len.
 */
#define pqremove(s, tree, top) \
{\
    top = s->heap[SMALLEST]; \
    s->heap[SMALLEST] = s->heap[s->heap_len--]; \
    pqdownheap(s, tree, SMALLEST); \
}

/* ===========================================================================
 * Compares to subtrees, using the tree depth as tie breaker when
 * the subtrees have equal frequency. This minimizes the worst case length.
 */
#define smaller(tree, n, m, depth) \
   (tree[n].Freq < tree[m].Freq || \
   (tree[n].Freq == tree[m].Freq && depth[n] <= depth[m]))

/* ===========================================================================
 * Restore the heap property by moving down the tree starting at node k,
 * exchanging a node with the smallest of its two sons if necessary, stopping
 * when the heap property is re-established (each father smaller than its
 * two sons).
 */
static void pqdownheap( deflate_state *s, ct_data *tree, int k )
{
    int v = s->heap[k];
    int j = k << 1;  /* left son of k */
    while (j <= s->heap_len) {
        /* Set j to the smallest of the two sons: */
        if (j < s->heap_len &&
            smaller(tree, s->heap[j+1], s->heap[j], s->depth)) {
            j++;
        }
        /* Exit if v is smaller than both sons */
        if (smaller(tree, v, s->heap[j], s->depth)) break;

        /* Exchange v with the smallest son */
        s->heap[k] = s->heap[j];  k = j;

        /* And continue down the tree, setting j to the left son of k */
        j <<= 1;
    }
    s->heap[k] = v;
}

/* ===========================================================================
 * Compute the optimal bit lengths for a tree and update the total bit length
 * for the current block.
 * IN assertion: the fields freq and dad are set, heap[heap_max] and
 *    above are the tree nodes sorted by increasing frequency.
 * OUT assertions: the field len is set to the optimal bit length, the
 *     array bl_count contains the frequencies for each bit length.
 *     The length opt_len is updated; static_len is also updated if stree is
 *     not null.
 */
static void gen_bitlen( deflate_state *s, tree_desc *desc )
{
    ct_data *tree        = desc->dyn_tree;
    int max_code         = desc->max_code;
    const ct_data *stree = desc->stat_desc->static_tree;
    const intf *extra    = desc->stat_desc->extra_bits;
    int base             = desc->stat_desc->extra_base;
    int max_length       = desc->stat_desc->max_length;
    int h;              /* heap index */
    int n, m;           /* iterate over the tree elements */
    int bits;           /* bit length */
    int xbits;          /* extra bits */
    ush f;              /* frequency */
    int overflow = 0;   /* number of elements with bit length too large */

    for (bits = 0; bits <= MAX_BITS; bits++) s->bl_count[bits] = 0;

    /* In a first pass, compute the optimal bit lengths (which may
     * overflow in the case of the bit length tree).
     */
    tree[s->heap[s->heap_max]].Len = 0; /* root of the heap */

    for (h = s->heap_max+1; h < HEAP_SIZE; h++) {
        n = s->heap[h];
        bits = tree[tree[n].Dad].Len + 1;
        if (bits > max_length) bits = max_length, overflow++;
        tree[n].Len = (ush)bits;
        /* We overwrite tree[n].Dad which is no longer needed */

        if (n > max_code) continue; /* not a leaf node */

        s->bl_count[bits]++;
        xbits = 0;
        if (n >= base) xbits = extra[n-base];
        f = tree[n].Freq;
        s->opt_len += (ulg)f * (unsigned)(bits + xbits);
        if (stree) s->static_len += (ulg)f * (unsigned)(stree[n].Len + xbits);
    }
    if (overflow == 0) return;

    Tracev((stderr,"\nbit length overflow\n"));
    /* This happens for example on obj2 and pic of the Calgary corpus */

    /* Find the first bit length which could increase: */
    do {
        bits = max_length-1;
        while (s->bl_count[bits] == 0) bits--;
        s->bl_count[bits]--;      /* move one leaf down the tree */
        s->bl_count[bits+1] += 2; /* move one overflow item as its brother */
        s->bl_count[max_length]--;
        /* The brother of the overflow item also moves one step up,
         * but this does not affect bl_count[max_length]
         */
        overflow -= 2;
    } while (overflow > 0);

    /* Now recompute all bit lengths, scanning in increasing frequency.
     * h is still equal to HEAP_SIZE. (It is simpler to reconstruct all
     * lengths instead of fixing only the wrong ones. This idea is taken
     * from 'ar' written by Haruhiko Okumura.)
     */
    for (bits = max_length; bits != 0; bits--) {
        n = s->bl_count[bits];
        while (n != 0) {
            m = s->heap[--h];
            if (m > max_code) continue;
            if ((unsigned) tree[m].Len != (unsigned) bits) {
                Tracev((stderr,"code %d bits %d->%d\n", m, tree[m].Len, bits));
                s->opt_len += ((ulg)bits - tree[m].Len) * tree[m].Freq;
                tree[m].Len = (ush)bits;
            }
            n--;
        }
    }
}

/* ===========================================================================
 * Generate the codes for a given tree and bit counts (which need not be
 * optimal).
 * IN assertion: the array bl_count contains the bit length statistics for
 * the given tree and the field len is set for all tree elements.
 * OUT assertion: the field code is set for all tree elements of non
 *     zero code length.
 */
static void gen_codes( ct_data *tree, int max_code, ushf *bl_count )
{
    ush next_code[MAX_BITS+1]; /* next code value for each bit length */
    unsigned code = 0;         /* running code value */
    int bits;                  /* bit index */
    int n;                     /* code index */

    /* The distribution counts are first used to generate the code values
     * without bit reversal.
     */
    for (bits = 1; bits <= MAX_BITS; bits++) {
        code = (code + bl_count[bits-1]) << 1;
        next_code[bits] = (ush)code;
    }
    /* Check that the bit counts in bl_count are consistent. The last code
     * must be all ones.
     */
    Assert (code + bl_count[MAX_BITS]-1 == (1<<MAX_BITS)-1,
            "inconsistent bit counts");
    Tracev((stderr,"\ngen_codes: max_code %d ", max_code));

    for (n = 0;  n <= max_code; n++) {
        int len = tree[n].Len;
        if (len == 0) continue;
        /* Now reverse the bits */
        tree[n].Code = (ush)bi_reverse(next_code[len]++, len);

        Tracecv(tree != static_ltree, (stderr,"\nn %3d %c l %2d c %4x (%x) ",
             n, (isgraph(n) ? n : ' '), len, tree[n].Code, next_code[len]-1));
    }
}

/* ===========================================================================
 * Construct one Huffman tree and assigns the code bit strings and lengths.
 * Update the total bit length for the current block.
 * IN assertion: the field freq is set for all tree elements.
 * OUT assertions: the fields len and code are set to the optimal bit length
 *     and corresponding code. The length opt_len is updated; static_len is
 *     also updated if stree is not null. The field max_code is set.
 */
static void build_tree( deflate_state *s, tree_desc *desc )
{
    ct_data *tree         = desc->dyn_tree;
    const ct_data *stree  = desc->stat_desc->static_tree;
    int elems             = desc->stat_desc->elems;
    int n, m;          /* iterate over heap elements */
    int max_code = -1; /* largest code with non zero frequency */
    int node;          /* new node being created */

    /* Construct the initial heap, with least frequent element in
     * heap[SMALLEST]. The sons of heap[n] are heap[2*n] and heap[2*n+1].
     * heap[0] is not used.
     */
    s->heap_len = 0, s->heap_max = HEAP_SIZE;

    for (n = 0; n < elems; n++) {
        if (tree[n].Freq != 0) {
            s->heap[++(s->heap_len)] = max_code = n;
            s->depth[n] = 0;
        } else {
            tree[n].Len = 0;
        }
    }

    /* The pkzip format requires that at least one distance code exists,
     * and that at least one bit should be sent even if there is only one
     * possible code. So to avoid special checks later on we force at least
     * two codes of non zero frequency.
     */
    while (s->heap_len < 2) {
        node = s->heap[++(s->heap_len)] = (max_code < 2 ? ++max_code : 0);
        tree[node].Freq = 1;
        s->depth[node] = 0;
        s->opt_len--; if (stree) s->static_len -= stree[node].Len;
        /* node is 0 or 1 so it does not have extra bits */
    }
    desc->max_code = max_code;

    /* The elements heap[heap_len/2+1 .. heap_len] are leaves of the tree,
     * establish sub-heaps of increasing lengths:
     */
    for (n = s->heap_len/2; n >= 1; n--) pqdownheap(s, tree, n);

    /* Construct the Huffman tree by repeatedly combining the least two
     * frequent nodes.
     */
    node = elems;              /* next internal node of the tree */
    do {
        pqremove(s, tree, n);  /* n = node of least frequency */
        m = s->heap[SMALLEST]; /* m = node of next least frequency */

        s->heap[--(s->heap_max)] = n; /* keep the nodes sorted by frequency */
        s->heap[--(s->heap_max)] = m;

        /* Create a new node father of n and m */
        tree[node].Freq = tree[n].Freq + tree[m].Freq;
        s->depth[node] = (uch)((s->depth[n] >= s->depth[m] ?
                                s->depth[n] : s->depth[m]) + 1);
        tree[n].Dad = tree[m].Dad = (ush)node;
#ifdef DUMP_BL_TREE
        if (tree == s->bl_tree) {
            fprintf(stderr,"\nnode %d(%d), sons %d(%d) %d(%d)",
                    node, tree[node].Freq, n, tree[n].Freq, m, tree[m].Freq);
        }
#endif
        /* and insert the new node in the heap */
        s->heap[SMALLEST] = node++;
        pqdownheap(s, tree, SMALLEST);

    } while (s->heap_len >= 2);

    s->heap[--(s->heap_max)] = s->heap[SMALLEST];

    /* At this point, the fields freq and dad are set. We can now
     * generate the bit lengths.
     */
    gen_bitlen(s, (tree_desc *)desc);

    /* The field len is now set, we can generate the bit codes */
    gen_codes ((ct_data *)tree, max_code, s->bl_count);
}

/* ===========================================================================
 * Scan a literal or distance tree to determine the frequencies of the codes
 * in the bit length tree.
 */
static void scan_tree( deflate_state *s, ct_data *tree, int max_code )
{
    int n;                     /* iterates over all tree elements */
    int prevlen = -1;          /* last emitted length */
    int curlen;                /* length of current code */
    int nextlen = tree[0].Len; /* length of next code */
    int count = 0;             /* repeat count of the current code */
    int max_count = 7;         /* max repeat count */
    int min_count = 4;         /* min repeat count */

    if (nextlen == 0) max_count = 138, min_count = 3;
    tree[max_code+1].Len = (ush)0xffff; /* guard */

    for (n = 0; n <= max_code; n++) {
        curlen = nextlen; nextlen = tree[n+1].Len;
        if (++count < max_count && curlen == nextlen) {
            continue;
        } else if (count < min_count) {
            s->bl_tree[curlen].Freq += count;
        } else if (curlen != 0) {
            if (curlen != prevlen) s->bl_tree[curlen].Freq++;
            s->bl_tree[REP_3_6].Freq++;
        } else if (count <= 10) {
            s->bl_tree[REPZ_3_10].Freq++;
        } else {
            s->bl_tree[REPZ_11_138].Freq++;
        }
        count = 0; prevlen = curlen;
        if (nextlen == 0) {
            max_count = 138, min_count = 3;
        } else if (curlen == nextlen) {
            max_count = 6, min_count = 3;
        } else {
            max_count = 7, min_count = 4;
        }
    }
}

/* ===========================================================================
 * Send a literal or distance tree in compressed form, using the codes in
 * bl_tree.
 */
static void send_tree( deflate_state *s, ct_data *tree, int max_code )
{
    int n;                     /* iterates over all tree elements */
    int prevlen = -1;          /* last emitted length */
    int curlen;                /* length of current code */
    int nextlen = tree[0].Len; /* length of next code */
    int count = 0;             /* repeat count of the current code */
    int max_count = 7;         /* max repeat count */
    int min_count = 4;         /* min repeat count */

    /* tree[max_code+1].Len = -1; */  /* guard already set */
    if (nextlen == 0) max_count = 138, min_count = 3;

    for (n = 0; n <= max_code; n++) {
        curlen = nextlen; nextlen = tree[n+1].Len;
        if (++count < max_count && curlen == nextlen) {
            continue;
        } else if (count < min_count) {
            do { send_code(s, curlen, s->bl_tree); } while (--count != 0);

        } else if (curlen != 0) {
            if (curlen != prevlen) {
                send_code(s, curlen, s->bl_tree); count--;
            }
            Assert(count >= 3 && count <= 6, " 3_6?");
            send_code(s, REP_3_6, s->bl_tree); send_bits(s, count-3, 2);

        } else if (count <= 10) {
            send_code(s, REPZ_3_10, s->bl_tree); send_bits(s, count-3, 3);

        } else {
            send_code(s, REPZ_11_138, s->bl_tree); send_bits(s, count-11, 7);
        }
        count = 0; prevlen = curlen;
        if (nextlen == 0) {
            max_count = 138, min_count = 3;
        } else if (curlen == nextlen) {
            max_count = 6, min_count = 3;
        } else {
            max_count = 7, min_count = 4;
        }
    }
}

/* ===========================================================================
 * Construct the Huffman tree for the bit lengths and return the index in
 * bl_order of the last bit length code to send.
 */
static int build_bl_tree( deflate_state *s )
{
    int max_blindex;  /* index of last bit length code of non zero freq */

    /* Determine the bit length frequencies for literal and distance trees */
    scan_tree(s, (ct_data *)s->dyn_ltree, s->l_desc.max_code);
    scan_tree(s, (ct_data *)s->dyn_dtree, s->d_desc.max_code);

    /* Build the bit length tree: */
    build_tree(s, (tree_desc *)(&(s->bl_desc)));
    /* opt_len now includes the length of the tree representations, except
     * the lengths of the bit lengths codes and the 5+5+4 bits for the counts.
     */

    /* Determine the number of bit length codes to send. The pkzip format
     * requires that at least 4 bit length codes be sent. (appnote.txt says
     * 3 but the actual value used is 4.)
     */
    for (max_blindex = BL_CODES-1; max_blindex >= 3; max_blindex--) {
        if (s->bl_tree[bl_order[max_blindex]].Len != 0) break;
    }
    /* Update opt_len to include the bit length tree and counts */
    s->opt_len += 3*((ulg)max_blindex+1) + 5+5+4;
    Tracev((stderr, "\ndyn trees: dyn %ld, stat %ld",
            s->opt_len, s->static_len));

    return max_blindex;
}

/* ===========================================================================
 * Send the header for a block using dynamic Huffman trees: the counts, the
 * lengths of the bit length codes, the literal tree and the distance tree.
 * IN assertion: lcodes >= 257, dcodes >= 1, blcodes >= 4.
 */
static void send_all_trees( deflate_state *s, int lcodes, int dcodes, int blcodes )
{
    int rank;                    /* index in bl_order */

    Assert (lcodes >= 257 && dcodes >= 1 && blcodes >= 4, "not enough codes");
    Assert (lcodes <= L_CODES && dcodes <= D_CODES && blcodes <= BL_CODES,
            "too many codes");
    Tracev((stderr, "\nbl counts: "));
    send_bits(s, lcodes-257, 5); /* not +255 as stated in appnote.txt */
    send_bits(s, dcodes-1,   5);
    send_bits(s, blcodes-4,  4); /* not -3 as stated in appnote.txt */
    for (rank = 0; rank < blcodes; rank++) {
        Tracev((stderr, "\nbl code %2d ", bl_order[rank]));
        send_bits(s, s->bl_tree[bl_order[rank]].Len, 3);
    }
    Tracev((stderr, "\nbl tree: sent %ld", s->bits_sent));

    send_tree(s, (ct_data *)s->dyn_ltree, lcodes-1); /* literal tree */
    Tracev((stderr, "\nlit tree: sent %ld", s->bits_sent));

    send_tree(s, (ct_data *)s->dyn_dtree, dcodes-1); /* distance tree */
    Tracev((stderr, "\ndist tree: sent %ld", s->bits_sent));
}

/* ===========================================================================
 * Send a stored block
 */
static void _tr_stored_block( deflate_state *s, charf *buf, ulg stored_len, int last )
{
    send_bits(s, (STORED_BLOCK<<1)+last, 3);    /* send block type */
    bi_windup(s);        /* align on byte boundary */
    put_short(s, (ush)stored_len);
    put_short(s, (ush)~stored_len);
    zmemcpy(s->pending_buf + s->pending, (Bytef *)buf, stored_len);
    s->pending += stored_len;
#ifdef ZLIB_DEBUG
    s->compressed_len = (s->compressed_len + 3 + 7) & (ulg)~7L;
    s->compressed_len += (stored_len + 4) << 3;
    s->bits_sent += 2*16;
    s->bits_sent += stored_len<<3;
#endif
}

/* ===========================================================================
 * Flush the bits in the bit buffer to pending output (leaves at most 7 bits)
 */
static void _tr_flush_bits( deflate_state *s )
{
    bi_flush(s);
}

/* ===========================================================================
 * Send one empty static block to give enough lookahead for inflate.
 * This takes 10 bits, of which 7 may remain in the bit buffer.
 */
static void _tr_align( deflate_state *s )
{
    send_bits(s, STATIC_TREES<<1, 3);
    send_code(s, END_BLOCK, static_ltree);
#ifdef ZLIB_DEBUG
    s->compressed_len += 10L; /* 3 for block type, 7 for EOB */
#endif
    bi_flush(s);
}

/* ===========================================================================
 * Determine the best encoding for the current block: dynamic trees, static
 * trees or store, and write out the encoded block.
 */
static void _tr_flush_block( deflate_state *s, charf *buf, ulg stored_len, int last )
{
    ulg opt_lenb, static_lenb; /* opt_len and static_len in bytes */
    int max_blindex = 0;  /* index of last bit length code of non zero freq */

    /* Build the Huffman trees unless a stored block is forced */
    if (s->level > 0) {

        /* Check if the file is binary or text */
        if (s->strm->data_type == Z_UNKNOWN)
            s->strm->data_type = detect_data_type(s);

        /* Construct the literal and distance trees */
        build_tree(s, (tree_desc *)(&(s->l_desc)));
        Tracev((stderr, "\nlit data: dyn %ld, stat %ld", s->opt_len,
                s->static_len));

        build_tree(s, (tree_desc *)(&(s->d_desc)));
        Tracev((stderr, "\ndist data: dyn %ld, stat %ld", s->opt_len,
                s->static_len));
        /* At this point, opt_len and static_len are the total bit lengths of
         * the compressed block data, excluding the tree representations.
         */

        /* Build the bit length tree for the above two trees, and get the index
         * in bl_order of the last bit length code to send.
         */
        max_blindex = build_bl_tree(s);

        /* Determine the best encoding. Compute the block lengths in bytes. */
        opt_lenb = (s->opt_len+3+7)>>3;
        static_lenb = (s->static_len+3+7)>>3;

        Tracev((stderr, "\nopt %lu(%lu) stat %lu(%lu) stored %lu lit %u ",
                opt_lenb, s->opt_len, static_lenb, s->static_len, stored_len,
                s->last_lit));

        if (static_lenb <= opt_lenb) opt_lenb = static_lenb;

    } else {
        Assert(buf != (char*)0, "lost buf");
        opt_lenb = static_lenb = stored_len + 5; /* force a stored block */
    }

#ifdef FORCE_STORED
    if (buf != (char*)0) { /* force stored block */
#else
    if (stored_len+4 <= opt_lenb && buf != (char*)0) {
                       /* 4: two words for the lengths */
#endif
        /* The test buf != NULL is only necessary if LIT_BUFSIZE > WSIZE.
         * Otherwise we can't have processed more than WSIZE input bytes since
         * the last block flush, because compression would have been
         * successful. If LIT_BUFSIZE <= WSIZE, it is never too late to
         * transform a block into a stored block.
         */
        _tr_stored_block(s, buf, stored_len, last);

#ifdef FORCE_STATIC
    } else if (static_lenb >= 0) { /* force static trees */
#else
    } else if (s->strategy == Z_FIXED || static_lenb == opt_lenb) {
#endif
        send_bits(s, (STATIC_TREES<<1)+last, 3);
        compress_block(s, (const ct_data *)static_ltree,
                       (const ct_data *)static_dtree);
#ifdef ZLIB_DEBUG
        s->compressed_len += 3 + s->static_len;
#endif
    } else {
        send_bits(s, (DYN_TREES<<1)+last, 3);
        send_all_trees(s, s->l_desc.max_code+1, s->d_desc.max_code+1,
                       max_blindex+1);
        compress_block(s, (const ct_data *)s->dyn_ltree,
                       (const ct_data *)s->dyn_dtree);
#ifdef ZLIB_DEBUG
        s->compressed_len += 3 + s->opt_len;
#endif
    }
    Assert (s->compressed_len == s->bits_sent, "bad compressed size");
    /* The above check is made mod 2^32, for files larger than 512 MB
     * and uLong implemented on 32 bits.
     */
    init_block(s);

    if (last) {
        bi_windup(s);
#ifdef ZLIB_DEBUG
        s->compressed_len += 7;  /* align on byte boundary */
#endif
    }
    Tracev((stderr,"\ncomprlen %lu(%lu) ", s->compressed_len>>3,
           s->compressed_len-7*last));
}

const char deflate_copyright[] =
   " deflate 1.2.11 Copyright 1995-2017 Jean-loup Gailly and Mark Adler ";
/*
  If you use the zlib library in a product, an acknowledgment is welcome
  in the documentation of your product. If for some reason you cannot
  include such an acknowledgment, I would appreciate that you keep this
  copyright string in the executable of your product.
 */

/* ===========================================================================
 *  Function prototypes.
 */
typedef enum {
    need_more,      /* block not completed, need more input or more output */
    block_done,     /* block flush performed */
    finish_started, /* finish started, need only more output at next deflate */
    finish_done     /* finish done, accept no more input or output */
} block_state;

typedef block_state (*compress_func)(deflate_state *s, int flush);
/* Compression function. Returns the block state after the call. */

static int deflateReset(z_streamp strm);
static block_state deflate_stored(deflate_state *s, int flush);
static block_state deflate_fast(deflate_state *s, int flush);
static block_state deflate_slow(deflate_state *s, int flush);
static block_state deflate_rle(deflate_state *s, int flush);
static block_state deflate_huff(deflate_state *s, int flush);
static void lm_init(deflate_state *s);

/* ===========================================================================
 * Local data
 */

#define NIL 0
/* Tail of hash chains */

#ifndef TOO_FAR
#  define TOO_FAR 4096
#endif
/* Matches of length 3 are discarded if their distance exceeds TOO_FAR */

/* Values for max_lazy_match, good_match and max_chain_length, depending on
 * the desired pack level (0..9). The values given below have been tuned to
 * exclude worst case performance for pathological files. Better values may be
 * found for specific files.
 */
typedef struct config_s {
   ush good_length; /* reduce lazy search above this match length */
   ush max_lazy;    /* do not perform lazy search above this match length */
   ush nice_length; /* quit search above this match length */
   ush max_chain;
   compress_func func;
} config;

static const config configuration_table[10] = {
/*      good lazy nice chain */
/* 0 */ {0,    0,  0,    0, deflate_stored},  /* store only */
/* 1 */ {4,    4,  8,    4, deflate_fast}, /* max speed, no lazy matches */
/* 2 */ {4,    5, 16,    8, deflate_fast},
/* 3 */ {4,    6, 32,   32, deflate_fast},

/* 4 */ {4,    4, 16,   16, deflate_slow},  /* lazy matches */
/* 5 */ {8,   16, 32,   32, deflate_slow},
/* 6 */ {8,   16, 128, 128, deflate_slow},
/* 7 */ {8,   32, 128, 256, deflate_slow},
/* 8 */ {32, 128, 258, 1024, deflate_slow},
/* 9 */ {32, 258, 258, 4096, deflate_slow}}; /* max compression */

/* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4
 * For deflate_fast() (levels <= 3) good is ignored and lazy has a different
 * meaning.
 */

/* rank Z_BLOCK between Z_NO_FLUSH and Z_PARTIAL_FLUSH */
#define RANK(f) (((f) * 2) - ((f) > 4 ? 9 : 0))

/* ===========================================================================
 * Update a hash value with the given input byte
 * IN  assertion: all calls to UPDATE_HASH are made with consecutive input
 *    characters, so that a running hash key can be computed from the previous
 *    key instead of complete recalculation each time.
 */
#define UPDATE_HASH(s,h,c) (h = (((h)<<s->hash_shift) ^ (c)) & s->hash_mask)


/* ===========================================================================
 * Insert string str in the dictionary and set match_head to the previous head
 * of the hash chain (the most recent string with same hash key). Return
 * the previous length of the hash chain.
 * If this file is compiled with -DFASTEST, the compression level is forced
 * to 1, and no hash chains are maintained.
 * IN  assertion: all calls to INSERT_STRING are made with consecutive input
 *    characters and the first MIN_MATCH bytes of str are valid (except for
 *    the last MIN_MATCH-1 bytes of the input file).
 */
#ifdef FASTEST
#define INSERT_STRING(s, str, match_head) \
   (UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \
    match_head = s->head[s->ins_h], \
    s->head[s->ins_h] = (Pos)(str))
#else
#define INSERT_STRING(s, str, match_head) \
   (UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \
    match_head = s->prev[(str) & s->w_mask] = s->head[s->ins_h], \
    s->head[s->ins_h] = (Pos)(str))
#endif

/* ===========================================================================
 * Initialize the hash table (avoiding 64K overflow for 16 bit systems).
 * prev[] will be initialized on the fly.
 */
#define CLEAR_HASH(s) \
    s->head[s->hash_size-1] = NIL; \
    zmemzero((Bytef *)s->head, (unsigned)(s->hash_size-1)*sizeof(*s->head));

/* ===========================================================================
 * Slide the hash table when sliding the window down (could be avoided with 32
 * bit values at the expense of memory usage). We slide even when level == 0 to
 * keep the hash table consistent if we switch back to level > 0 later.
 */
static void slide_hash( deflate_state *s )
{
    unsigned n, m;
    Posf *p;
    uInt wsize = s->w_size;

    n = s->hash_size;
    p = &s->head[n];
    do {
        m = *--p;
        *p = (Pos)(m >= wsize ? m - wsize : NIL);
    } while (--n);
    n = wsize;
#ifndef FASTEST
    p = &s->prev[n];
    do {
        m = *--p;
        *p = (Pos)(m >= wsize ? m - wsize : NIL);
        /* If n is not on any hash chain, prev[n] is garbage but
         * its value will never be used.
         */
    } while (--n);
#endif
}

/* ========================================================================= */
int deflateInit( z_streamp strm, int level )
{
    return deflateInit2(strm, level, Z_DEFLATED, MAX_WBITS, DEF_MEM_LEVEL, Z_DEFAULT_STRATEGY);
    /* To do: ignore strm->next_in if we use it as window */
}

/* ========================================================================= */
int deflateInit2( z_streamp strm, int level, int method, int windowBits, int memLevel, int strategy )
{
    deflate_state *s;
    int wrap = 1;
    ushf *overlay;
    /* We overlay pending_buf and d_buf+l_buf. This works since the average
     * output size for (length,distance) codes is <= 24 bits.
     */

    strm->msg = Z_NULL;
#ifdef FASTEST
    if (level != 0) level = 1;
#else
    if (level == Z_DEFAULT_COMPRESSION) level = 6;
#endif

    if (windowBits < 0) { /* suppress zlib wrapper */
        wrap = 0;
        windowBits = -windowBits;
    }
#ifdef GZIP
    else if (windowBits > 15) {
        wrap = 2;       /* write gzip wrapper instead */
        windowBits -= 16;
    }
#endif
    if (memLevel < 1 || memLevel > MAX_MEM_LEVEL || method != Z_DEFLATED ||
        windowBits < 8 || windowBits > 15 || level < 0 || level > 9 ||
        strategy < 0 || strategy > Z_FIXED || (windowBits == 8 && wrap != 1)) {
        return Z_STREAM_ERROR;
    }
    if (windowBits == 8) windowBits = 9;  /* until 256-byte window bug fixed */
    s = (deflate_state *) ZALLOC(strm, 1, sizeof(deflate_state));
    if (s == Z_NULL) return Z_MEM_ERROR;
    strm->state = (struct internal_state FAR *)s;
    s->strm = strm;
    s->status = INIT_STATE;     /* to pass state test in deflateReset() */

    s->wrap = wrap;
    s->gzhead = Z_NULL;
    s->w_bits = (uInt)windowBits;
    s->w_size = 1 << s->w_bits;
    s->w_mask = s->w_size - 1;

    s->hash_bits = (uInt)memLevel + 7;
    s->hash_size = 1 << s->hash_bits;
    s->hash_mask = s->hash_size - 1;
    s->hash_shift =  ((s->hash_bits+MIN_MATCH-1)/MIN_MATCH);

    s->window = (Bytef *) ZALLOC(strm, s->w_size, 2*sizeof(Byte));
    s->prev   = (Posf *)  ZALLOC(strm, s->w_size, sizeof(Pos));
    s->head   = (Posf *)  ZALLOC(strm, s->hash_size, sizeof(Pos));

    s->high_water = 0;      /* nothing written to s->window yet */

    s->lit_bufsize = 1 << (memLevel + 6); /* 16K elements by default */

    overlay = (ushf *) ZALLOC(strm, s->lit_bufsize, sizeof(ush)+2);
    s->pending_buf = (uchf *) overlay;
    s->pending_buf_size = (ulg)s->lit_bufsize * (sizeof(ush)+2L);

    if (s->window == Z_NULL || s->prev == Z_NULL || s->head == Z_NULL ||
        s->pending_buf == Z_NULL) {
        s->status = FINISH_STATE;
        strm->msg = ERR_MSG(Z_MEM_ERROR);
        deflateEnd (strm);
        return Z_MEM_ERROR;
    }
    s->d_buf = overlay + s->lit_bufsize/sizeof(ush);
    s->l_buf = s->pending_buf + (1+sizeof(ush))*s->lit_bufsize;

    s->level = level;
    s->strategy = strategy;
    s->method = (Byte)method;

    return deflateReset(strm);
}

/* =========================================================================
 * Check for a valid deflate stream state. Return 0 if ok, 1 if not.
 */
static int deflateStateCheck( z_streamp strm )
{
    deflate_state *s;
    if (strm == Z_NULL ||
        strm->zalloc == (alloc_func)0 || strm->zfree == (free_func)0)
        return 1;
    s = strm->state;
    if (s == Z_NULL || s->strm != strm || (s->status != INIT_STATE &&
#ifdef GZIP
                                           s->status != GZIP_STATE &&
#endif
                                           s->status != EXTRA_STATE &&
                                           s->status != NAME_STATE &&
                                           s->status != COMMENT_STATE &&
                                           s->status != HCRC_STATE &&
                                           s->status != BUSY_STATE &&
                                           s->status != FINISH_STATE))
        return 1;
    return 0;
}

/* ========================================================================= */
static int deflateResetKeep( z_streamp strm )
{
    deflate_state *s;

    if (deflateStateCheck(strm)) {
        return Z_STREAM_ERROR;
    }

    strm->total_in = strm->total_out = 0;
    strm->msg = Z_NULL; /* use zfree if we ever allocate msg dynamically */
    strm->data_type = Z_UNKNOWN;

    s = (deflate_state *)strm->state;
    s->pending = 0;
    s->pending_out = s->pending_buf;

    if (s->wrap < 0) {
        s->wrap = -s->wrap; /* was made negative by deflate(..., Z_FINISH); */
    }
    s->status =
#ifdef GZIP
        s->wrap == 2 ? GZIP_STATE :
#endif
        s->wrap ? INIT_STATE : BUSY_STATE;
    strm->adler =
#ifdef GZIP
        s->wrap == 2 ? crc32(0L, Z_NULL, 0) :
#endif
        adler32(0L, Z_NULL, 0);
    s->last_flush = Z_NO_FLUSH;

    _tr_init(s);

    return Z_OK;
}

/* ========================================================================= */
static int deflateReset( z_streamp strm )
{
    int ret;

    ret = deflateResetKeep(strm);
    if (ret == Z_OK)
        lm_init(strm->state);
    return ret;
}

/* =========================================================================
 * Put a short in the pending buffer. The 16-bit value is put in MSB order.
 * IN assertion: the stream state is correct and there is enough room in
 * pending_buf.
 */
static void putShortMSB( deflate_state *s, uInt b )
{
    put_byte(s, (Byte)(b >> 8));
    put_byte(s, (Byte)(b & 0xff));
}

/* =========================================================================
 * Flush as much pending output as possible. All deflate() output, except for
 * some deflate_stored() output, goes through this function so some
 * applications may wish to modify it to avoid allocating a large
 * strm->next_out buffer and copying into it. (See also read_buf()).
 */
static void flush_pending( z_streamp strm )
{
    unsigned len;
    deflate_state *s = strm->state;

    _tr_flush_bits(s);
    len = s->pending;
    if (len > strm->avail_out) len = strm->avail_out;
    if (len == 0) return;

    zmemcpy(strm->next_out, s->pending_out, len);
    strm->next_out  += len;
    s->pending_out  += len;
    strm->total_out += len;
    strm->avail_out -= len;
    s->pending      -= len;
    if (s->pending == 0) {
        s->pending_out = s->pending_buf;
    }
}

/* ===========================================================================
 * Update the header CRC with the bytes s->pending_buf[beg..s->pending - 1].
 */
#define HCRC_UPDATE(beg) \
    do { \
        if (s->gzhead->hcrc && s->pending > (beg)) \
            strm->adler = crc32(strm->adler, s->pending_buf + (beg), \
                                s->pending - (beg)); \
    } while (0)

/* ========================================================================= */
int deflate( z_streamp strm, int flush )
{
    int old_flush; /* value of flush param for previous deflate call */
    deflate_state *s;

    if (deflateStateCheck(strm) || flush > Z_BLOCK || flush < 0) {
        return Z_STREAM_ERROR;
    }
    s = strm->state;

    if (strm->next_out == Z_NULL ||
        (strm->avail_in != 0 && strm->next_in == Z_NULL) ||
        (s->status == FINISH_STATE && flush != Z_FINISH)) {
        ERR_RETURN(strm, Z_STREAM_ERROR);
    }
    if (strm->avail_out == 0) ERR_RETURN(strm, Z_BUF_ERROR);

    old_flush = s->last_flush;
    s->last_flush = flush;

    /* Flush as much pending output as possible */
    if (s->pending != 0) {
        flush_pending(strm);
        if (strm->avail_out == 0) {
            /* Since avail_out is 0, deflate will be called again with
             * more output space, but possibly with both pending and
             * avail_in equal to zero. There won't be anything to do,
             * but this is not an error situation so make sure we
             * return OK instead of BUF_ERROR at next call of deflate:
             */
            s->last_flush = -1;
            return Z_OK;
        }

    /* Make sure there is something to do and avoid duplicate consecutive
     * flushes. For repeated and useless calls with Z_FINISH, we keep
     * returning Z_STREAM_END instead of Z_BUF_ERROR.
     */
    } else if (strm->avail_in == 0 && RANK(flush) <= RANK(old_flush) &&
               flush != Z_FINISH) {
        ERR_RETURN(strm, Z_BUF_ERROR);
    }

    /* User must not provide more input after the first FINISH: */
    if (s->status == FINISH_STATE && strm->avail_in != 0) {
        ERR_RETURN(strm, Z_BUF_ERROR);
    }

    /* Write the header */
    if (s->status == INIT_STATE) {
        /* zlib header */
        uInt header = (Z_DEFLATED + ((s->w_bits-8)<<4)) << 8;
        uInt level_flags;

        if (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2)
            level_flags = 0;
        else if (s->level < 6)
            level_flags = 1;
        else if (s->level == 6)
            level_flags = 2;
        else
            level_flags = 3;
        header |= (level_flags << 6);
        if (s->strstart != 0) header |= PRESET_DICT;
        header += 31 - (header % 31);

        putShortMSB(s, header);

        /* Save the adler32 of the preset dictionary: */
        if (s->strstart != 0) {
            putShortMSB(s, (uInt)(strm->adler >> 16));
            putShortMSB(s, (uInt)(strm->adler & 0xffff));
        }
        strm->adler = adler32(0L, Z_NULL, 0);
        s->status = BUSY_STATE;

        /* Compression must start with an empty pending buffer */
        flush_pending(strm);
        if (s->pending != 0) {
            s->last_flush = -1;
            return Z_OK;
        }
    }
#ifdef GZIP
    if (s->status == GZIP_STATE) {
        /* gzip header */
        strm->adler = crc32(0L, Z_NULL, 0);
        put_byte(s, 31);
        put_byte(s, 139);
        put_byte(s, 8);
        if (s->gzhead == Z_NULL) {
            put_byte(s, 0);
            put_byte(s, 0);
            put_byte(s, 0);
            put_byte(s, 0);
            put_byte(s, 0);
            put_byte(s, s->level == 9 ? 2 :
                     (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ?
                      4 : 0));
            put_byte(s, OS_CODE);
            s->status = BUSY_STATE;

            /* Compression must start with an empty pending buffer */
            flush_pending(strm);
            if (s->pending != 0) {
                s->last_flush = -1;
                return Z_OK;
            }
        }
        else {
            put_byte(s, (s->gzhead->text ? 1 : 0) +
                     (s->gzhead->hcrc ? 2 : 0) +
                     (s->gzhead->extra == Z_NULL ? 0 : 4) +
                     (s->gzhead->name == Z_NULL ? 0 : 8) +
                     (s->gzhead->comment == Z_NULL ? 0 : 16)
                     );
            put_byte(s, (Byte)(s->gzhead->time & 0xff));
            put_byte(s, (Byte)((s->gzhead->time >> 8) & 0xff));
            put_byte(s, (Byte)((s->gzhead->time >> 16) & 0xff));
            put_byte(s, (Byte)((s->gzhead->time >> 24) & 0xff));
            put_byte(s, s->level == 9 ? 2 :
                     (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ?
                      4 : 0));
            put_byte(s, s->gzhead->os & 0xff);
            if (s->gzhead->extra != Z_NULL) {
                put_byte(s, s->gzhead->extra_len & 0xff);
                put_byte(s, (s->gzhead->extra_len >> 8) & 0xff);
            }
            if (s->gzhead->hcrc)
                strm->adler = crc32(strm->adler, s->pending_buf,
                                    s->pending);
            s->gzindex = 0;
            s->status = EXTRA_STATE;
        }
    }
    if (s->status == EXTRA_STATE) {
        if (s->gzhead->extra != Z_NULL) {
            ulg beg = s->pending;   /* start of bytes to update crc */
            uInt left = (s->gzhead->extra_len & 0xffff) - s->gzindex;
            while (s->pending + left > s->pending_buf_size) {
                uInt copy = s->pending_buf_size - s->pending;
                zmemcpy(s->pending_buf + s->pending,
                        s->gzhead->extra + s->gzindex, copy);
                s->pending = s->pending_buf_size;
                HCRC_UPDATE(beg);
                s->gzindex += copy;
                flush_pending(strm);
                if (s->pending != 0) {
                    s->last_flush = -1;
                    return Z_OK;
                }
                beg = 0;
                left -= copy;
            }
            zmemcpy(s->pending_buf + s->pending,
                    s->gzhead->extra + s->gzindex, left);
            s->pending += left;
            HCRC_UPDATE(beg);
            s->gzindex = 0;
        }
        s->status = NAME_STATE;
    }
    if (s->status == NAME_STATE) {
        if (s->gzhead->name != Z_NULL) {
            ulg beg = s->pending;   /* start of bytes to update crc */
            int val;
            do {
                if (s->pending == s->pending_buf_size) {
                    HCRC_UPDATE(beg);
                    flush_pending(strm);
                    if (s->pending != 0) {
                        s->last_flush = -1;
                        return Z_OK;
                    }
                    beg = 0;
                }
                val = s->gzhead->name[s->gzindex++];
                put_byte(s, val);
            } while (val != 0);
            HCRC_UPDATE(beg);
            s->gzindex = 0;
        }
        s->status = COMMENT_STATE;
    }
    if (s->status == COMMENT_STATE) {
        if (s->gzhead->comment != Z_NULL) {
            ulg beg = s->pending;   /* start of bytes to update crc */
            int val;
            do {
                if (s->pending == s->pending_buf_size) {
                    HCRC_UPDATE(beg);
                    flush_pending(strm);
                    if (s->pending != 0) {
                        s->last_flush = -1;
                        return Z_OK;
                    }
                    beg = 0;
                }
                val = s->gzhead->comment[s->gzindex++];
                put_byte(s, val);
            } while (val != 0);
            HCRC_UPDATE(beg);
        }
        s->status = HCRC_STATE;
    }
    if (s->status == HCRC_STATE) {
        if (s->gzhead->hcrc) {
            if (s->pending + 2 > s->pending_buf_size) {
                flush_pending(strm);
                if (s->pending != 0) {
                    s->last_flush = -1;
                    return Z_OK;
                }
            }
            put_byte(s, (Byte)(strm->adler & 0xff));
            put_byte(s, (Byte)((strm->adler >> 8) & 0xff));
            strm->adler = crc32(0L, Z_NULL, 0);
        }
        s->status = BUSY_STATE;

        /* Compression must start with an empty pending buffer */
        flush_pending(strm);
        if (s->pending != 0) {
            s->last_flush = -1;
            return Z_OK;
        }
    }
#endif

    /* Start a new block or continue the current one.
     */
    if (strm->avail_in != 0 || s->lookahead != 0 ||
        (flush != Z_NO_FLUSH && s->status != FINISH_STATE)) {
        block_state bstate;

        bstate = s->level == 0 ? deflate_stored(s, flush) :
                 s->strategy == Z_HUFFMAN_ONLY ? deflate_huff(s, flush) :
                 s->strategy == Z_RLE ? deflate_rle(s, flush) :
                 (*(configuration_table[s->level].func))(s, flush);

        if (bstate == finish_started || bstate == finish_done) {
            s->status = FINISH_STATE;
        }
        if (bstate == need_more || bstate == finish_started) {
            if (strm->avail_out == 0) {
                s->last_flush = -1; /* avoid BUF_ERROR next call, see above */
            }
            return Z_OK;
            /* If flush != Z_NO_FLUSH && avail_out == 0, the next call
             * of deflate should use the same flush parameter to make sure
             * that the flush is complete. So we don't have to output an
             * empty block here, this will be done at next call. This also
             * ensures that for a very small output buffer, we emit at most
             * one empty block.
             */
        }
        if (bstate == block_done) {
            if (flush == Z_PARTIAL_FLUSH) {
                _tr_align(s);
            } else if (flush != Z_BLOCK) { /* FULL_FLUSH or SYNC_FLUSH */
                _tr_stored_block(s, (char*)0, 0L, 0);
                /* For a full flush, this empty block will be recognized
                 * as a special marker by inflate_sync().
                 */
                if (flush == Z_FULL_FLUSH) {
                    CLEAR_HASH(s);             /* forget history */
                    if (s->lookahead == 0) {
                        s->strstart = 0;
                        s->block_start = 0L;
                        s->insert = 0;
                    }
                }
            }
            flush_pending(strm);
            if (strm->avail_out == 0) {
              s->last_flush = -1; /* avoid BUF_ERROR at next call, see above */
              return Z_OK;
            }
        }
    }

    if (flush != Z_FINISH) return Z_OK;
    if (s->wrap <= 0) return Z_STREAM_END;

    /* Write the trailer */
#ifdef GZIP
    if (s->wrap == 2) {
        put_byte(s, (Byte)(strm->adler & 0xff));
        put_byte(s, (Byte)((strm->adler >> 8) & 0xff));
        put_byte(s, (Byte)((strm->adler >> 16) & 0xff));
        put_byte(s, (Byte)((strm->adler >> 24) & 0xff));
        put_byte(s, (Byte)(strm->total_in & 0xff));
        put_byte(s, (Byte)((strm->total_in >> 8) & 0xff));
        put_byte(s, (Byte)((strm->total_in >> 16) & 0xff));
        put_byte(s, (Byte)((strm->total_in >> 24) & 0xff));
    }
    else
#endif
    {
        putShortMSB(s, (uInt)(strm->adler >> 16));
        putShortMSB(s, (uInt)(strm->adler & 0xffff));
    }
    flush_pending(strm);
    /* If avail_out is zero, the application will call deflate again
     * to flush the rest.
     */
    if (s->wrap > 0) s->wrap = -s->wrap; /* write the trailer only once! */
    return s->pending != 0 ? Z_OK : Z_STREAM_END;
}

/* ========================================================================= */
int deflateEnd( z_streamp strm )
{
    int status;

    if (deflateStateCheck(strm)) return Z_STREAM_ERROR;

    status = strm->state->status;

    /* Deallocate in reverse order of allocations: */
    TRY_FREE(strm, strm->state->pending_buf);
    TRY_FREE(strm, strm->state->head);
    TRY_FREE(strm, strm->state->prev);
    TRY_FREE(strm, strm->state->window);

    ZFREE(strm, strm->state);
    strm->state = Z_NULL;

    return status == BUSY_STATE ? Z_DATA_ERROR : Z_OK;
}

/* ===========================================================================
 * Read a new buffer from the current input stream, update the adler32
 * and total number of bytes read.  All deflate() input goes through
 * this function so some applications may wish to modify it to avoid
 * allocating a large strm->next_in buffer and copying from it.
 * (See also flush_pending()).
 */
static unsigned read_buf( z_streamp strm, Bytef *buf, unsigned size )
{
    unsigned len = strm->avail_in;

    if (len > size) len = size;
    if (len == 0) return 0;

    strm->avail_in  -= len;

    zmemcpy(buf, strm->next_in, len);
    if (strm->state->wrap == 1) {
        strm->adler = adler32(strm->adler, buf, len);
    }
#ifdef GZIP
    else if (strm->state->wrap == 2) {
        strm->adler = crc32(strm->adler, buf, len);
    }
#endif
    strm->next_in  += len;
    strm->total_in += len;

    return len;
}

/* ===========================================================================
 * Initialize the "longest match" routines for a new zlib stream
 */
static void lm_init( deflate_state *s )
{
    s->window_size = (ulg)2L*s->w_size;

    CLEAR_HASH(s);

    /* Set the default configuration parameters:
     */
    s->max_lazy_match   = configuration_table[s->level].max_lazy;
    s->good_match       = configuration_table[s->level].good_length;
    s->nice_match       = configuration_table[s->level].nice_length;
    s->max_chain_length = configuration_table[s->level].max_chain;

    s->strstart = 0;
    s->block_start = 0L;
    s->lookahead = 0;
    s->insert = 0;
    s->match_length = s->prev_length = MIN_MATCH-1;
    s->match_available = 0;
    s->ins_h = 0;
}

/* ===========================================================================
 * Set match_start to the longest match starting at the given string and
 * return its length. Matches shorter or equal to prev_length are discarded,
 * in which case the result is equal to prev_length and match_start is
 * garbage.
 * IN assertions: cur_match is the head of the hash chain for the current
 *   string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1
 * OUT assertion: the match length is not greater than s->lookahead.
 */
static uInt longest_match( deflate_state *s, IPos cur_match )
{
    unsigned chain_length = s->max_chain_length;/* max hash chain length */
    register Bytef *scan = s->window + s->strstart; /* current string */
    register Bytef *match;                      /* matched string */
    register int len;                           /* length of current match */
    int best_len = (int)s->prev_length;         /* best match length so far */
    int nice_match = s->nice_match;             /* stop if match long enough */
    IPos limit = s->strstart > (IPos)MAX_DIST(s) ?
        s->strstart - (IPos)MAX_DIST(s) : NIL;
    /* Stop when cur_match becomes <= limit. To simplify the code,
     * we prevent matches with the string of window index 0.
     */
    Posf *prev = s->prev;
    uInt wmask = s->w_mask;

#ifdef UNALIGNED_OK
    /* Compare two bytes at a time. Note: this is not always beneficial.
     * Try with and without -DUNALIGNED_OK to check.
     */
    register Bytef *strend = s->window + s->strstart + MAX_MATCH - 1;
    register ush scan_start = *(ushf*)scan;
    register ush scan_end   = *(ushf*)(scan+best_len-1);
#else
    register Bytef *strend = s->window + s->strstart + MAX_MATCH;
    register Byte scan_end1  = scan[best_len-1];
    register Byte scan_end   = scan[best_len];
#endif

    /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16.
     * It is easy to get rid of this optimization if necessary.
     */
    Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever");

    /* Do not waste too much time if we already have a good match: */
    if (s->prev_length >= s->good_match) {
        chain_length >>= 2;
    }
    /* Do not look for matches beyond the end of the input. This is necessary
     * to make deflate deterministic.
     */
    if ((uInt)nice_match > s->lookahead) nice_match = (int)s->lookahead;

    Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead");

    do {
        Assert(cur_match < s->strstart, "no future");
        match = s->window + cur_match;

        /* Skip to next match if the match length cannot increase
         * or if the match length is less than 2.  Note that the checks below
         * for insufficient lookahead only occur occasionally for performance
         * reasons.  Therefore uninitialized memory will be accessed, and
         * conditional jumps will be made that depend on those values.
         * However the length of the match is limited to the lookahead, so
         * the output of deflate is not affected by the uninitialized values.
         */
#if (defined(UNALIGNED_OK) && MAX_MATCH == 258)
        /* This code assumes sizeof(unsigned short) == 2. Do not use
         * UNALIGNED_OK if your compiler uses a different size.
         */
        if (*(ushf*)(match+best_len-1) != scan_end ||
            *(ushf*)match != scan_start) continue;

        /* It is not necessary to compare scan[2] and match[2] since they are
         * always equal when the other bytes match, given that the hash keys
         * are equal and that HASH_BITS >= 8. Compare 2 bytes at a time at
         * strstart+3, +5, ... up to strstart+257. We check for insufficient
         * lookahead only every 4th comparison; the 128th check will be made
         * at strstart+257. If MAX_MATCH-2 is not a multiple of 8, it is
         * necessary to put more guard bytes at the end of the window, or
         * to check more often for insufficient lookahead.
         */
        Assert(scan[2] == match[2], "scan[2]?");
        scan++, match++;
        do {
        } while (*(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
                 *(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
                 *(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
                 *(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
                 scan < strend);
        /* The funny "do {}" generates better code on most compilers */

        /* Here, scan <= window+strstart+257 */
        Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");
        if (*scan == *match) scan++;

        len = (MAX_MATCH - 1) - (int)(strend-scan);
        scan = strend - (MAX_MATCH-1);

#else /* UNALIGNED_OK */

        if (match[best_len]   != scan_end  ||
            match[best_len-1] != scan_end1 ||
            *match            != *scan     ||
            *++match          != scan[1])      continue;

        /* The check at best_len-1 can be removed because it will be made
         * again later. (This heuristic is not always a win.)
         * It is not necessary to compare scan[2] and match[2] since they
         * are always equal when the other bytes match, given that
         * the hash keys are equal and that HASH_BITS >= 8.
         */
        scan += 2, match++;
        Assert(*scan == *match, "match[2]?");

        /* We check for insufficient lookahead only every 8th comparison;
         * the 256th check will be made at strstart+258.
         */
        do {
        } while (*++scan == *++match && *++scan == *++match &&
                 *++scan == *++match && *++scan == *++match &&
                 *++scan == *++match && *++scan == *++match &&
                 *++scan == *++match && *++scan == *++match &&
                 scan < strend);

        Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");

        len = MAX_MATCH - (int)(strend - scan);
        scan = strend - MAX_MATCH;

#endif /* UNALIGNED_OK */

        if (len > best_len) {
            s->match_start = cur_match;
            best_len = len;
            if (len >= nice_match) break;
#ifdef UNALIGNED_OK
            scan_end = *(ushf*)(scan+best_len-1);
#else
            scan_end1  = scan[best_len-1];
            scan_end   = scan[best_len];
#endif
        }
    } while ((cur_match = prev[cur_match & wmask]) > limit
             && --chain_length != 0);

    if ((uInt)best_len <= s->lookahead) return (uInt)best_len;
    return s->lookahead;
}

#define check_match(s, start, match, length)

/* ===========================================================================
 * Fill the window when the lookahead becomes insufficient.
 * Updates strstart and lookahead.
 *
 * IN assertion: lookahead < MIN_LOOKAHEAD
 * OUT assertions: strstart <= window_size-MIN_LOOKAHEAD
 *    At least one byte has been read, or avail_in == 0; reads are
 *    performed for at least two bytes (required for the zip translate_eol
 *    option -- not supported here).
 */
static void fill_window( deflate_state *s )
{
    unsigned n;
    unsigned more;    /* Amount of free space at the end of the window. */
    uInt wsize = s->w_size;

    Assert(s->lookahead < MIN_LOOKAHEAD, "already enough lookahead");

    do {
        more = (unsigned)(s->window_size -(ulg)s->lookahead -(ulg)s->strstart);

        /* Deal with !@#$% 64K limit: */
        if (sizeof(int) <= 2) {
            if (more == 0 && s->strstart == 0 && s->lookahead == 0) {
                more = wsize;

            } else if (more == (unsigned)(-1)) {
                /* Very unlikely, but possible on 16 bit machine if
                 * strstart == 0 && lookahead == 1 (input done a byte at time)
                 */
                more--;
            }
        }

        /* If the window is almost full and there is insufficient lookahead,
         * move the upper half to the lower one to make room in the upper half.
         */
        if (s->strstart >= wsize+MAX_DIST(s)) {

            zmemcpy(s->window, s->window+wsize, (unsigned)wsize - more);
            s->match_start -= wsize;
            s->strstart    -= wsize; /* we now have strstart >= MAX_DIST */
            s->block_start -= (long) wsize;
            slide_hash(s);
            more += wsize;
        }
        if (s->strm->avail_in == 0) break;

        /* If there was no sliding:
         *    strstart <= WSIZE+MAX_DIST-1 && lookahead <= MIN_LOOKAHEAD - 1 &&
         *    more == window_size - lookahead - strstart
         * => more >= window_size - (MIN_LOOKAHEAD-1 + WSIZE + MAX_DIST-1)
         * => more >= window_size - 2*WSIZE + 2
         * In the BIG_MEM or MMAP case (not yet supported),
         *   window_size == input_size + MIN_LOOKAHEAD  &&
         *   strstart + s->lookahead <= input_size => more >= MIN_LOOKAHEAD.
         * Otherwise, window_size == 2*WSIZE so more >= 2.
         * If there was sliding, more >= WSIZE. So in all cases, more >= 2.
         */
        Assert(more >= 2, "more < 2");

        n = read_buf(s->strm, s->window + s->strstart + s->lookahead, more);
        s->lookahead += n;

        /* Initialize the hash value now that we have some input: */
        if (s->lookahead + s->insert >= MIN_MATCH) {
            uInt str = s->strstart - s->insert;
            s->ins_h = s->window[str];
            UPDATE_HASH(s, s->ins_h, s->window[str + 1]);
#if MIN_MATCH != 3
            Call UPDATE_HASH() MIN_MATCH-3 more times
#endif
            while (s->insert) {
                UPDATE_HASH(s, s->ins_h, s->window[str + MIN_MATCH-1]);
#ifndef FASTEST
                s->prev[str & s->w_mask] = s->head[s->ins_h];
#endif
                s->head[s->ins_h] = (Pos)str;
                str++;
                s->insert--;
                if (s->lookahead + s->insert < MIN_MATCH)
                    break;
            }
        }
        /* If the whole input has less than MIN_MATCH bytes, ins_h is garbage,
         * but this is not important since only literal bytes will be emitted.
         */

    } while (s->lookahead < MIN_LOOKAHEAD && s->strm->avail_in != 0);

    /* If the WIN_INIT bytes after the end of the current data have never been
     * written, then zero those bytes in order to avoid memory check reports of
     * the use of uninitialized (or uninitialised as Julian writes) bytes by
     * the longest match routines.  Update the high water mark for the next
     * time through here.  WIN_INIT is set to MAX_MATCH since the longest match
     * routines allow scanning to strstart + MAX_MATCH, ignoring lookahead.
     */
    if (s->high_water < s->window_size) {
        ulg curr = s->strstart + (ulg)(s->lookahead);
        ulg init;

        if (s->high_water < curr) {
            /* Previous high water mark below current data -- zero WIN_INIT
             * bytes or up to end of window, whichever is less.
             */
            init = s->window_size - curr;
            if (init > WIN_INIT)
                init = WIN_INIT;
            zmemzero(s->window + curr, (unsigned)init);
            s->high_water = curr + init;
        }
        else if (s->high_water < (ulg)curr + WIN_INIT) {
            /* High water mark at or above current data, but below current data
             * plus WIN_INIT -- zero out to current data plus WIN_INIT, or up
             * to end of window, whichever is less.
             */
            init = (ulg)curr + WIN_INIT - s->high_water;
            if (init > s->window_size - s->high_water)
                init = s->window_size - s->high_water;
            zmemzero(s->window + s->high_water, (unsigned)init);
            s->high_water += init;
        }
    }

    Assert((ulg)s->strstart <= s->window_size - MIN_LOOKAHEAD,
           "not enough room for search");
}

/* ===========================================================================
 * Flush the current block, with given end-of-file flag.
 * IN assertion: strstart is set to the end of the current match.
 */
#define FLUSH_BLOCK_ONLY(s, last) { \
   _tr_flush_block(s, (s->block_start >= 0L ? \
                   (charf *)&s->window[(unsigned)s->block_start] : \
                   (charf *)Z_NULL), \
                (ulg)((long)s->strstart - s->block_start), \
                (last)); \
   s->block_start = s->strstart; \
   flush_pending(s->strm); \
   Tracev((stderr,"[FLUSH]")); \
}

/* Same but force premature exit if necessary. */
#define FLUSH_BLOCK(s, last) { \
   FLUSH_BLOCK_ONLY(s, last); \
   if (s->strm->avail_out == 0) return (last) ? finish_started : need_more; \
}

/* Maximum stored block length in deflate format (not including header). */
#define MAX_STORED 65535

/* Minimum of a and b. */
#define MIN(a, b) ((a) > (b) ? (b) : (a))

/* ===========================================================================
 * Copy without compression as much as possible from the input stream, return
 * the current block state.
 *
 * In case deflateParams() is used to later switch to a non-zero compression
 * level, s->matches (otherwise unused when storing) keeps track of the number
 * of hash table slides to perform. If s->matches is 1, then one hash table
 * slide will be done when switching. If s->matches is 2, the maximum value
 * allowed here, then the hash table will be cleared, since two or more slides
 * is the same as a clear.
 *
 * deflate_stored() is written to minimize the number of times an input byte is
 * copied. It is most efficient with large input and output buffers, which
 * maximizes the opportunites to have a single copy from next_in to next_out.
 */
static block_state deflate_stored( deflate_state *s, int flush )
{
    /* Smallest worthy block size when not flushing or finishing. By default
     * this is 32K. This can be as small as 507 bytes for memLevel == 1. For
     * large input and output buffers, the stored block size will be larger.
     */
    unsigned min_block = MIN(s->pending_buf_size - 5, s->w_size);

    /* Copy as many min_block or larger stored blocks directly to next_out as
     * possible. If flushing, copy the remaining available input to next_out as
     * stored blocks, if there is enough space.
     */
    unsigned len, left, have, last = 0;
    unsigned used = s->strm->avail_in;
    do {
        /* Set len to the maximum size block that we can copy directly with the
         * available input data and output space. Set left to how much of that
         * would be copied from what's left in the window.
         */
        len = MAX_STORED;       /* maximum deflate stored block length */
        have = (s->bi_valid + 42) >> 3;         /* number of header bytes */
        if (s->strm->avail_out < have)          /* need room for header */
            break;
            /* maximum stored block length that will fit in avail_out: */
        have = s->strm->avail_out - have;
        left = s->strstart - s->block_start;    /* bytes left in window */
        if (len > (ulg)left + s->strm->avail_in)
            len = left + s->strm->avail_in;     /* limit len to the input */
        if (len > have)
            len = have;                         /* limit len to the output */

        /* If the stored block would be less than min_block in length, or if
         * unable to copy all of the available input when flushing, then try
         * copying to the window and the pending buffer instead. Also don't
         * write an empty block when flushing -- deflate() does that.
         */
        if (len < min_block && ((len == 0 && flush != Z_FINISH) ||
                                flush == Z_NO_FLUSH ||
                                len != left + s->strm->avail_in))
            break;

        /* Make a dummy stored block in pending to get the header bytes,
         * including any pending bits. This also updates the debugging counts.
         */
        last = flush == Z_FINISH && len == left + s->strm->avail_in ? 1 : 0;
        _tr_stored_block(s, (char *)0, 0L, last);

        /* Replace the lengths in the dummy stored block with len. */
        s->pending_buf[s->pending - 4] = len;
        s->pending_buf[s->pending - 3] = len >> 8;
        s->pending_buf[s->pending - 2] = ~len;
        s->pending_buf[s->pending - 1] = ~len >> 8;

        /* Write the stored block header bytes. */
        flush_pending(s->strm);

#ifdef ZLIB_DEBUG
        /* Update debugging counts for the data about to be copied. */
        s->compressed_len += len << 3;
        s->bits_sent += len << 3;
#endif

        /* Copy uncompressed bytes from the window to next_out. */
        if (left) {
            if (left > len)
                left = len;
            zmemcpy(s->strm->next_out, s->window + s->block_start, left);
            s->strm->next_out += left;
            s->strm->avail_out -= left;
            s->strm->total_out += left;
            s->block_start += left;
            len -= left;
        }

        /* Copy uncompressed bytes directly from next_in to next_out, updating
         * the check value.
         */
        if (len) {
            read_buf(s->strm, s->strm->next_out, len);
            s->strm->next_out += len;
            s->strm->avail_out -= len;
            s->strm->total_out += len;
        }
    } while (last == 0);

    /* Update the sliding window with the last s->w_size bytes of the copied
     * data, or append all of the copied data to the existing window if less
     * than s->w_size bytes were copied. Also update the number of bytes to
     * insert in the hash tables, in the event that deflateParams() switches to
     * a non-zero compression level.
     */
    used -= s->strm->avail_in;      /* number of input bytes directly copied */
    if (used) {
        /* If any input was used, then no unused input remains in the window,
         * therefore s->block_start == s->strstart.
         */
        if (used >= s->w_size) {    /* supplant the previous history */
            s->matches = 2;         /* clear hash */
            zmemcpy(s->window, s->strm->next_in - s->w_size, s->w_size);
            s->strstart = s->w_size;
        }
        else {
            if (s->window_size - s->strstart <= used) {
                /* Slide the window down. */
                s->strstart -= s->w_size;
                zmemcpy(s->window, s->window + s->w_size, s->strstart);
                if (s->matches < 2)
                    s->matches++;   /* add a pending slide_hash() */
            }
            zmemcpy(s->window + s->strstart, s->strm->next_in - used, used);
            s->strstart += used;
        }
        s->block_start = s->strstart;
        s->insert += MIN(used, s->w_size - s->insert);
    }
    if (s->high_water < s->strstart)
        s->high_water = s->strstart;

    /* If the last block was written to next_out, then done. */
    if (last)
        return finish_done;

    /* If flushing and all input has been consumed, then done. */
    if (flush != Z_NO_FLUSH && flush != Z_FINISH &&
        s->strm->avail_in == 0 && (long)s->strstart == s->block_start)
        return block_done;

    /* Fill the window with any remaining input. */
    have = s->window_size - s->strstart - 1;
    if (s->strm->avail_in > have && s->block_start >= (long)s->w_size) {
        /* Slide the window down. */
        s->block_start -= s->w_size;
        s->strstart -= s->w_size;
        zmemcpy(s->window, s->window + s->w_size, s->strstart);
        if (s->matches < 2)
            s->matches++;           /* add a pending slide_hash() */
        have += s->w_size;          /* more space now */
    }
    if (have > s->strm->avail_in)
        have = s->strm->avail_in;
    if (have) {
        read_buf(s->strm, s->window + s->strstart, have);
        s->strstart += have;
    }
    if (s->high_water < s->strstart)
        s->high_water = s->strstart;

    /* There was not enough avail_out to write a complete worthy or flushed
     * stored block to next_out. Write a stored block to pending instead, if we
     * have enough input for a worthy block, or if flushing and there is enough
     * room for the remaining input as a stored block in the pending buffer.
     */
    have = (s->bi_valid + 42) >> 3;         /* number of header bytes */
        /* maximum stored block length that will fit in pending: */
    have = MIN(s->pending_buf_size - have, MAX_STORED);
    min_block = MIN(have, s->w_size);
    left = s->strstart - s->block_start;
    if (left >= min_block ||
        ((left || flush == Z_FINISH) && flush != Z_NO_FLUSH &&
         s->strm->avail_in == 0 && left <= have)) {
        len = MIN(left, have);
        last = flush == Z_FINISH && s->strm->avail_in == 0 &&
               len == left ? 1 : 0;
        _tr_stored_block(s, (charf *)s->window + s->block_start, len, last);
        s->block_start += len;
        flush_pending(s->strm);
    }

    /* We've done all we can with the available input and output. */
    return last ? finish_started : need_more;
}

/* ===========================================================================
 * Compress as much as possible from the input stream, return the current
 * block state.
 * This function does not perform lazy evaluation of matches and inserts
 * new strings in the dictionary only for unmatched strings or for short
 * matches. It is used only for the fast compression options.
 */
static block_state deflate_fast( deflate_state *s, int flush )
{
    IPos hash_head;       /* head of the hash chain */
    int bflush;           /* set if current block must be flushed */

    for (;;) {
        /* Make sure that we always have enough lookahead, except
         * at the end of the input file. We need MAX_MATCH bytes
         * for the next match, plus MIN_MATCH bytes to insert the
         * string following the next match.
         */
        if (s->lookahead < MIN_LOOKAHEAD) {
            fill_window(s);
            if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) {
                return need_more;
            }
            if (s->lookahead == 0) break; /* flush the current block */
        }

        /* Insert the string window[strstart .. strstart+2] in the
         * dictionary, and set hash_head to the head of the hash chain:
         */
        hash_head = NIL;
        if (s->lookahead >= MIN_MATCH) {
            INSERT_STRING(s, s->strstart, hash_head);
        }

        /* Find the longest match, discarding those <= prev_length.
         * At this point we have always match_length < MIN_MATCH
         */
        if (hash_head != NIL && s->strstart - hash_head <= MAX_DIST(s)) {
            /* To simplify the code, we prevent matches with the string
             * of window index 0 (in particular we have to avoid a match
             * of the string with itself at the start of the input file).
             */
            s->match_length = longest_match (s, hash_head);
            /* longest_match() sets match_start */
        }
        if (s->match_length >= MIN_MATCH) {
            check_match(s, s->strstart, s->match_start, s->match_length);

            _tr_tally_dist(s, s->strstart - s->match_start,
                           s->match_length - MIN_MATCH, bflush);

            s->lookahead -= s->match_length;

            /* Insert new strings in the hash table only if the match length
             * is not too large. This saves time but degrades compression.
             */
#ifndef FASTEST
            if (s->match_length <= s->max_insert_length &&
                s->lookahead >= MIN_MATCH) {
                s->match_length--; /* string at strstart already in table */
                do {
                    s->strstart++;
                    INSERT_STRING(s, s->strstart, hash_head);
                    /* strstart never exceeds WSIZE-MAX_MATCH, so there are
                     * always MIN_MATCH bytes ahead.
                     */
                } while (--s->match_length != 0);
                s->strstart++;
            } else
#endif
            {
                s->strstart += s->match_length;
                s->match_length = 0;
                s->ins_h = s->window[s->strstart];
                UPDATE_HASH(s, s->ins_h, s->window[s->strstart+1]);
#if MIN_MATCH != 3
                Call UPDATE_HASH() MIN_MATCH-3 more times
#endif
                /* If lookahead < MIN_MATCH, ins_h is garbage, but it does not
                 * matter since it will be recomputed at next deflate call.
                 */
            }
        } else {
            /* No match, output a literal byte */
            Tracevv((stderr,"%c", s->window[s->strstart]));
            _tr_tally_lit (s, s->window[s->strstart], bflush);
            s->lookahead--;
            s->strstart++;
        }
        if (bflush) FLUSH_BLOCK(s, 0);
    }
    s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1;
    if (flush == Z_FINISH) {
        FLUSH_BLOCK(s, 1);
        return finish_done;
    }
    if (s->last_lit)
        FLUSH_BLOCK(s, 0);
    return block_done;
}

/* ===========================================================================
 * Same as above, but achieves better compression. We use a lazy
 * evaluation for matches: a match is finally adopted only if there is
 * no better match at the next window position.
 */
static block_state deflate_slow( deflate_state *s, int flush )
{
    IPos hash_head;          /* head of hash chain */
    int bflush;              /* set if current block must be flushed */

    /* Process the input block. */
    for (;;) {
        /* Make sure that we always have enough lookahead, except
         * at the end of the input file. We need MAX_MATCH bytes
         * for the next match, plus MIN_MATCH bytes to insert the
         * string following the next match.
         */
        if (s->lookahead < MIN_LOOKAHEAD) {
            fill_window(s);
            if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) {
                return need_more;
            }
            if (s->lookahead == 0) break; /* flush the current block */
        }

        /* Insert the string window[strstart .. strstart+2] in the
         * dictionary, and set hash_head to the head of the hash chain:
         */
        hash_head = NIL;
        if (s->lookahead >= MIN_MATCH) {
            INSERT_STRING(s, s->strstart, hash_head);
        }

        /* Find the longest match, discarding those <= prev_length.
         */
        s->prev_length = s->match_length, s->prev_match = s->match_start;
        s->match_length = MIN_MATCH-1;

        if (hash_head != NIL && s->prev_length < s->max_lazy_match &&
            s->strstart - hash_head <= MAX_DIST(s)) {
            /* To simplify the code, we prevent matches with the string
             * of window index 0 (in particular we have to avoid a match
             * of the string with itself at the start of the input file).
             */
            s->match_length = longest_match (s, hash_head);
            /* longest_match() sets match_start */

            if (s->match_length <= 5 && (s->strategy == Z_FILTERED
#if TOO_FAR <= 32767
                || (s->match_length == MIN_MATCH &&
                    s->strstart - s->match_start > TOO_FAR)
#endif
                )) {

                /* If prev_match is also MIN_MATCH, match_start is garbage
                 * but we will ignore the current match anyway.
                 */
                s->match_length = MIN_MATCH-1;
            }
        }
        /* If there was a match at the previous step and the current
         * match is not better, output the previous match:
         */
        if (s->prev_length >= MIN_MATCH && s->match_length <= s->prev_length) {
            uInt max_insert = s->strstart + s->lookahead - MIN_MATCH;
            /* Do not insert strings in hash table beyond this. */

            check_match(s, s->strstart-1, s->prev_match, s->prev_length);

            _tr_tally_dist(s, s->strstart -1 - s->prev_match,
                           s->prev_length - MIN_MATCH, bflush);

            /* Insert in hash table all strings up to the end of the match.
             * strstart-1 and strstart are already inserted. If there is not
             * enough lookahead, the last two strings are not inserted in
             * the hash table.
             */
            s->lookahead -= s->prev_length-1;
            s->prev_length -= 2;
            do {
                if (++s->strstart <= max_insert) {
                    INSERT_STRING(s, s->strstart, hash_head);
                }
            } while (--s->prev_length != 0);
            s->match_available = 0;
            s->match_length = MIN_MATCH-1;
            s->strstart++;

            if (bflush) FLUSH_BLOCK(s, 0);

        } else if (s->match_available) {
            /* If there was no match at the previous position, output a
             * single literal. If there was a match but the current match
             * is longer, truncate the previous match to a single literal.
             */
            Tracevv((stderr,"%c", s->window[s->strstart-1]));
            _tr_tally_lit(s, s->window[s->strstart-1], bflush);
            if (bflush) {
                FLUSH_BLOCK_ONLY(s, 0);
            }
            s->strstart++;
            s->lookahead--;
            if (s->strm->avail_out == 0) return need_more;
        } else {
            /* There is no previous match to compare with, wait for
             * the next step to decide.
             */
            s->match_available = 1;
            s->strstart++;
            s->lookahead--;
        }
    }
    Assert (flush != Z_NO_FLUSH, "no flush?");
    if (s->match_available) {
        Tracevv((stderr,"%c", s->window[s->strstart-1]));
        _tr_tally_lit(s, s->window[s->strstart-1], bflush);
        s->match_available = 0;
    }
    s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1;
    if (flush == Z_FINISH) {
        FLUSH_BLOCK(s, 1);
        return finish_done;
    }
    if (s->last_lit)
        FLUSH_BLOCK(s, 0);
    return block_done;
}

/* ===========================================================================
 * For Z_RLE, simply look for runs of bytes, generate matches only of distance
 * one.  Do not maintain a hash table.  (It will be regenerated if this run of
 * deflate switches away from Z_RLE.)
 */
static block_state deflate_rle( deflate_state *s, int flush )
{
    int bflush;             /* set if current block must be flushed */
    uInt prev;              /* byte at distance one to match */
    Bytef *scan, *strend;   /* scan goes up to strend for length of run */

    for (;;) {
        /* Make sure that we always have enough lookahead, except
         * at the end of the input file. We need MAX_MATCH bytes
         * for the longest run, plus one for the unrolled loop.
         */
        if (s->lookahead <= MAX_MATCH) {
            fill_window(s);
            if (s->lookahead <= MAX_MATCH && flush == Z_NO_FLUSH) {
                return need_more;
            }
            if (s->lookahead == 0) break; /* flush the current block */
        }

        /* See how many times the previous byte repeats */
        s->match_length = 0;
        if (s->lookahead >= MIN_MATCH && s->strstart > 0) {
            scan = s->window + s->strstart - 1;
            prev = *scan;
            if (prev == *++scan && prev == *++scan && prev == *++scan) {
                strend = s->window + s->strstart + MAX_MATCH;
                do {
                } while (prev == *++scan && prev == *++scan &&
                         prev == *++scan && prev == *++scan &&
                         prev == *++scan && prev == *++scan &&
                         prev == *++scan && prev == *++scan &&
                         scan < strend);
                s->match_length = MAX_MATCH - (uInt)(strend - scan);
                if (s->match_length > s->lookahead)
                    s->match_length = s->lookahead;
            }
            Assert(scan <= s->window+(uInt)(s->window_size-1), "wild scan");
        }

        /* Emit match if have run of MIN_MATCH or longer, else emit literal */
        if (s->match_length >= MIN_MATCH) {
            check_match(s, s->strstart, s->strstart - 1, s->match_length);

            _tr_tally_dist(s, 1, s->match_length - MIN_MATCH, bflush);

            s->lookahead -= s->match_length;
            s->strstart += s->match_length;
            s->match_length = 0;
        } else {
            /* No match, output a literal byte */
            Tracevv((stderr,"%c", s->window[s->strstart]));
            _tr_tally_lit (s, s->window[s->strstart], bflush);
            s->lookahead--;
            s->strstart++;
        }
        if (bflush) FLUSH_BLOCK(s, 0);
    }
    s->insert = 0;
    if (flush == Z_FINISH) {
        FLUSH_BLOCK(s, 1);
        return finish_done;
    }
    if (s->last_lit)
        FLUSH_BLOCK(s, 0);
    return block_done;
}

/* ===========================================================================
 * For Z_HUFFMAN_ONLY, do not look for matches.  Do not maintain a hash table.
 * (It will be regenerated if this run of deflate switches away from Huffman.)
 */
static block_state deflate_huff( deflate_state *s, int flush )
{
    int bflush;             /* set if current block must be flushed */

    for (;;) {
        /* Make sure that we have a literal to write. */
        if (s->lookahead == 0) {
            fill_window(s);
            if (s->lookahead == 0) {
                if (flush == Z_NO_FLUSH)
                    return need_more;
                break;      /* flush the current block */
            }
        }

        /* Output a literal byte */
        s->match_length = 0;
        Tracevv((stderr,"%c", s->window[s->strstart]));
        _tr_tally_lit (s, s->window[s->strstart], bflush);
        s->lookahead--;
        s->strstart++;
        if (bflush) FLUSH_BLOCK(s, 0);
    }
    s->insert = 0;
    if (flush == Z_FINISH) {
        FLUSH_BLOCK(s, 1);
        return finish_done;
    }
    if (s->last_lit)
        FLUSH_BLOCK(s, 0);
    return block_done;
}